Abstract

Sm-Nd garnet and U-Pb zircon ages for eclogite and granulite from the Breaksea Orthogneiss provide a detailed chronology for pluton emplacement and subsequent thermal history of the lower arc crust exposed in Fiordland, New Zealand. The 147Sm-143Nd ages for ∼1 cm garnet grains in eclogite yield a 108.2 ± 1.8 Ma (7 points) age and similar sized grains of garnet from granulite interlayered with eclogite yield a ca. 110.5 ± 1.6 Ma (8 points) age. Both samples retain sparse domains with older ages of 123–121 Ma. Distinct Ca, Lu, and Hf zoning in garnet indicate that eclogite and granulite cooled rapidly enough to negate significant diffusion. The Ca zoning is interpreted to indicate significant garnet recrystallization during the granulite facies event, ca. 110 Ma. The older garnet ages are indistinguishable from the oldest 206U/238Pb zircon ages, ca. 123 and 120 Ma, in granulite orthogneiss that yielded two age populations; these granulites have younger age populations of 111.1 ± 1.4 and 115.2 ± 1.3 Ma, respectively. Zircon from orthogneiss samples nearby yield single age populations indicating additional intrusions ca. 115 and late metamorphic zircon growth ca. 95 Ma. The zircon and garnet ages combined with pressure-temperature-time paths document magma intrusion into the lowermost arc crust, near isothermal exhumation of Breaksea rocks at ∼2.2 km/m.y. from ∼65 km to 40–45 km depths, followed by continued high heat flow with granulite facies metamorphism. The latter high temperatures were synchronous with granulite facies metamorphism in the adjacent Malaspina pluton, indicating that high-temperature metamorphism affected >600 km2 of lower crust in the continental magmatic arc.

The complex age results for U-Pb zircon and Sm-Nd garnet dating indicate the need for comprehensive data sets from multiple rocks for deciphering the intrusive and subsequent thermal history of the lower crust. The study detailed here clearly indicates that Sm-Nd garnet geochronology can provide useful ages for high-temperature rocks when large grains cool at rates of >10 °C/m.y. The geochronological results indicate that voluminous magmatism was closely followed by high-temperature metamorphism. This is a common phenomenon in the lower crust of magmatic arcs and a signature for high magmatic flux through the lower crust.

INTRODUCTION

The timing of magmatism and metamorphism is key to determining the fluxes of magma into the crust and the thermal structure of arc crust. The large pressure and temperature stability of zircon and garnet, their widespread occurrence, and their high closure temperatures for geochronology have led to common use of both minerals for dating events in the mid- to lower crust (e.g., Harley et al., 2007; Baxter and Scherer, 2013). High-temperature eclogite and granulite facies gneisses from the area around Breaksea Sound in Fiordland, New Zealand (Fig. 1), provide a natural laboratory for studying lower crustal magmatic arc root processes with integrated geochronological, structural, and petrological data. In these deepest exposed parts of the arc root, the metamorphic history provides a means for determining the nature and processes responsible for arc magmatism.

Eclogite and granulite in the Breaksea Orthogneiss are the deepest exposed rocks of the Fiordland magmatic arc. These rocks are proposed to have had the same high pressure and temperature history (De Paoli et al., 2009, 2012), yet geochronological data to elucidate the history are lacking. Complex structural relationships between the eclogite and granulite, which are deformed into domes and locally highly strained (e.g., Klepeis et al., 2016), lead to uncertainty about the primary intrusive relationships between these rocks. Clarke et al. (2013) describe interlayered eclogite and granulite from the Breaksea Orthogneiss that are interpreted as primary igneous cumulate layering. However, the extent of recrystallization is not certain and confirmation of age relationships between igneous and metamorphic processes is needed.

Although zircon is an excellent geochronometer, the interpretation of zircon ages in high-temperature rocks may be complicated by loss of radiogenic Pb and complex zircon growth during multiple igneous, metamorphic, and hydrothermal events. Lattice damage resulting from U decay and/or externally imposed strain may significantly enhance Pb mobility and loss, and zircon dissolution and/or reprecipitation (Cherniak and Watson, 2001). Sm and Nd isotopes in garnet can also provide a robust geochronometer appropriate for a wide range of metamorphic temperatures (e.g., Baxter and Scherer, 2013); however, mineral inclusions in garnet and diffusion of Sm and Nd at temperatures over 700 °C may pose problems (Prince et al., 2000; Scherer et al., 2000). In addition, both garnet and zircon may grow during igneous crystallization and metamorphism (e.g., Jagoutz and Kelemen, 2015), thus mixed populations may be present. Determining the age of magmatism and metamorphism in the lower arc crust exposed in Fiordland is subject to all of these complexities because metamorphic temperatures exceeded 800 °C (Oliver, 1977; Bradshaw, 1989; Stowell et al., 2014) and some of the garnet is likely to be igneous (Clarke et al., 2013). This study, which applies Sm-Nd garnet geochronology to high-temperature eclogite and granulite, utilizes large garnet grains, trace element concentrations in garnet, and acid leaching techniques to produce the best data and to evaluate the results.

This contribution presents new U-Pb zircon and Sm-Nd garnet isotope data, and pseudosection-based pressure-temperature-time (P-T-t) paths (see GSA Data Repository Item1 for all methods) in order to constrain the timing and P-T conditions of pluton emplacement, eclogite facies metamorphism, and later upper amphibolite to garnet granulite metamorphism in the deepest exposed part of the Fiordland magmatic arc, i.e., the Breaksea Orthogneiss from the entrance to Breaksea Sound (Figs. 1 and 2). Outcrop to microscopic textures, major and trace element mineral compositions, and compositional zoning are used with garnet and zircon geochronology to distinguish between igneous and metamorphic zircon and garnet growth in orthogneiss samples (Table 1). Observations at the outcrop and hand-sample scale, including primary igneous layering of coarse pyroxene-rich gabbro and crosscutting metamorphic fabrics that include amphibole and/or garnet, allow some garnet to be clearly identified as metamorphic. Garnet major element concentrations and zoning constrain the P-T conditions during metamorphism to those of garnet granulite conditions. Rare earth element (REE) concentrations in coexisting garnet and zircon (e.g., Hoskin and Schaltegger, 2003) indicate that garnet and zircon likely grew independently without significant influence from uptake by synchronous growth. These results indicate that (1) igneous and/or eclogite facies garnet and zircon growth was 123–119 Ma, (2) most of the garnet in the Breaksea Orthogneiss is metamorphic and grew ca. 110 Ma at 12–14 kbar granulite facies conditions, (3) two ca. 95 Ma metamorphic zircon ages indicate that the Breaksea and Resolution Orthogneisses, west of the Resolution Island shear zones, were at amphibolite facies temperatures ∼10 m.y. after garnet granulite metamorphism, (4) the new ages for garnet growth indicate that high-temperature eclogite facies metamorphism predates granulite conditions by ∼10 m.y., and (5) granulite garnet ages in the Breaksea Orthogneiss are younger than most garnet granulite metamorphism in the Malaspina pluton, indicating that granulite conditions may have lasted ∼2 m.y. longer in the Breaksea Sound area.

GEOLOGICAL SETTING

Fiordland is composed of an exhumed magmatic arc that formed on the eastern margin of Gondwana from the Carboniferous to the Early Cretaceous (Tulloch and Kimbrough, 2003; Klepeis et al., 2003; Mortimer et al., 1999, 2004). Two belts of plutons dominate the magmatic rocks: an inboard belt of 123–115 Ma high Sr/Y plutons known as the Western Fiordland Orthogneiss (WFO) (Mattinson et al., 1986; Allibone et al., 2009a; Schwartz et al., 2016) and an outboard belt of dominantly 147–136 Ma low Sr/Y plutons known as the Darran suite (Tulloch and Kimbrough, 2003; Schwartz et al., 2016). The WFO is dominated by monzonite, monzodiorite, and diorite of the Worsley, Misty, and Malaspina plutons, which form the lower crust of the magmatic arc from north to south, respectively (Gibson and Ireland, 1995; Allibone et al., 2009a). The WFO locally intruded and elsewhere was fault juxtaposed with metasedimentary and metaigneous rocks of the Paleozoic Takaka terrane (Turnbull et al., 2010). In the area from Deep Cove to Dusky Sound, dominantly metasedimentary rocks of the Takaka terrane have been named the Deep Cove Gneiss (Oliver, 1980; Gibson, 1982).

The lower arc crust around Breaksea Sound consists of the Malaspina pluton and the Resolution and Breaksea Orthogneisses, which are juxtaposed along the Resolution Island shear zone and the Straight River shear zone (King et al., 2008; Turnbull et al., 2010; Betka and Klepeis, 2013) (Fig. 2). The Malaspina pluton was emplaced during a high magma addition rate event between 118 and 115 Ma (Schwartz et al., 2017). In the Malaspina pluton, garnet granulite facies metamorphism is well documented to have reached ∼850 °C and 12–14 kbar between 113 and 111 Ma (Stowell et al., 2014). The rocks around Breaksea Sound include eclogite, indicating that this area is the deepest exposed part of the Cretaceous magmatic arc (De Paoli et al., 2009, 2012; Clarke et al., 2013). The Breaksea Orthogneiss is a composite unit with interlayered garnet granulite and eclogite. These layers have compositions that include garnet pyroxenite, monzogabbro, and monzodiorite. Although the eclogite to high-pressure granulite facies metamorphism in the Breaksea Orthogneiss is known to have peaked at ≥850 °C and ∼18 kbar (De Paoli et al., 2009), the timing of metamorphism is poorly constrained. In addition, complex mineral textures and mineral compositional data in these high-grade rocks have led to uncertainty about whether garnet and pyroxene crystallized directly from magma, forming eclogite layers, or grew later during metamorphism (Clarke et al., 2013).

RESULTS

Mineral Textures and Compositions

We collected 7 eclogite and 16 orthogneiss samples from the Breaksea Sound and Acheron Passage areas (Figs. 1 and 2) for this study. Of these samples, two eclogite and five orthogneiss samples were selected for detailed petrologic and geochronologic studies. Eclogite occurs as layers, lenses, and outcrop-scale domes within mafic orthogneiss (Fig. 3). The eclogite domes are wrapped by highly strained granulite, which locally contains lenses of amphibole and plagioclase ± garnet and pyroxene. Clots and grain aggregates of amphibole plus biotite are interpreted to result from replacement of garnet and pyroxene.

Breaksea eclogite on Wairaki, Hawea, and northern Resolution Islands occurs in domes with diameters that range from ∼10 m to ∼1 km (Fig. 2) interlayered with granulite. Aggregates of <1-cm-diameter garnet crystals commonly form lenses to 1 m in length. In addition, garnet locally occurs as isolated >5 cm euhedral megacrysts.

Eclogite 12NZ01

Eclogite sample 12NZ01b is a coarse-grained and massive to weakly layered gneiss collected from Wairaki Island (Fig. 2). The sample is dominated by garnet and pyroxene with amphibole and minor plagioclase (Fig. 4). Accessory phases include rutile, magnetite, clinozoisite, and scapolite. Eclogite sample 12NZ01a, also collected on Wairaki Island, is composed of garnet aggregates rimmed by amphibole and plagioclase all set in an omphacitic clinopyroxene groundmass with accessory ilmenite, rutile, and magnetite. Plagioclase and Na- rich amphibole locally form micrographic intergrowths, some of which form radiating textures around pyroxene and garnet.

Garnet occurs as <1 cm fractured subhedral to anhedral grains with rutile, and as irregularly shaped aggregates of subhedral 2–5-mm-diameter grains, which are commonly arranged around the larger crystals. Inclusions of pyroxene and rutile are common throughout the larger garnet grains. Rutile ranges from needles to rounded grains, both of which lack obvious preferred crystallographic orientations. Al, Si, Ca, Fe, Mg, Mn, K, and Na Kα X-ray maps for 12NZ01a show significant and near-ubiquitous zoning in garnet and minor zoning in pyroxene (Figs. 5 and 6). Garnet grains are strongly zoned (Fig. 6) with ∼2-mm-diameter low-Ca high-Mg cores (grossular, Grs = 0.16, almandine, Alm = 0.41, and pyrope, Prp = 0.39), 1–2-mm-wide high Ca mantles (Grs = 0.23, Alm = 0.41, and Prp = 0.35), and narrow <0.5 mm rims attributed to diffusion. Garnet compositions are similar to those described by Clarke et al. (2013) for samples obtained from near Mount Richards informally called Breaksea Tops and by De Paoli et al. (2009) from Breaksea Sound. However, garnet described here contains significantly higher grossular (Table 2A) than that described in De Paoli et al. (2009), and the garnet zoning from our samples with cores that have Prp = ∼40% and Grs = ∼17% to rims with Prp = 34% and Grs = 23% is more pronounced with broader rims. Manganese concentrations (spessartine, Sps = 0.01–0.02) are low and near constant from core to rim (Fig. 6).

Omphacitic clinopyroxene is fractured, subhedral to anhedral, and occurs as subrounded 1–2-mm-diameter grains defining 0.5–2-cm-thick gneissic layering. Compositional zoning is easily seen in thin section and X-ray maps as a change from low Ca and high Na in broad cores to high Ca and Mg, and low Na content in narrow rims (Table 2B). This is commonly accompanied by a change from cores with abundant exsolution lamellae to rims lacking obvious exsolution.

Amphibole occurs locally as masses of subequant grains with plagioclase, euhedral blue-green crystals in oriented growth with plagioclase, and as radiating prisms along grain boundaries of garnet and pyroxene. The amphibole is interpreted as post-eclogite facies based on its association with plagioclase and common intergrowths with plagioclase around garnet and pyroxene.

A few early plagioclase grains can be found as anhedral inclusions in garnet rims and within pyroxene. Late plagioclase is found in cracks of garnet and in grain boundary intergrowth with amphibole. Rutile is found as inclusions within major phases and in the matrix. Scapolite and clinozoisite are strictly found in the matrix filling in spaces between clinopyroxene and amphibole.

Breaksea Orthogneiss sample 12NZ01d is medium-grained biotite amphibole gneiss from the southern end of Wairaki Island (Fig. 2). The foliation is defined by the alignment of hornblende and biotite. Minor quartz is strongly recrystallized with sutured grain boundaries suggestive of recovery by grain boundary migration. Trace amounts of apatite, ilmenite, and zircon are present and rutile is locally found included in hornblende and plagioclase, but is absent in the matrix. Secondary anhedral titanite is locally intergrown with ilmenite.

Granulite 12NZ04

Granulite sample 12NZ04 is medium- to coarse-grained garnet pyroxene gneiss collected from the Breaksea Orthogneiss on Hawea Island (Fig. 2). Compositional layering is defined by coarse lenses of pyroxene and garnet interlayered with plagioclase-rich layers. Concordant leucosomes, interpreted as former melts, contain plagioclase and quartz with small garnet porphyroblasts. The sample is composed of plagioclase, garnet, and clinopyroxene with minor amounts of amphibole, biotite, and quartz. There are two distinct types of garnet, subhedral to euhedral garnet usually ∼1 cm in diameter, and smaller anhedral garnet grains with skeletal form. Some garnet crystals contain crystallographically orientated rutile needles indicating exsolution of Ti from the garnet lattice. Large garnet grains contain plagioclase (near rim) and pyroxene inclusions and numerous fractures containing biotite and plagioclase. Intergrowths of plagioclase, pyroxene, and minor amounts of amphibole and biotite surround most garnet. Accessory phases in the rock include rutile, apatite, and clinozoisite.

Al, Si, Ca, Fe, Mg, Mn, K, and Na Kα X-ray maps for sample 12NZ04b show little or no zoning in most minerals. However, the larger garnets in this sample are weakly zoned (Figs. 5 and 6). Most of the large garnet crystals have near-constant Ca, Mn, Mg, and Fe (Grs = 0.14, Sps = 0.02, Prp = 0.39, and Alm = 0.45). However, Fe increases and Mg decreases in an ∼400 µm zone at the rims of grains. The smaller anhedral garnet, pyroxene, and plagioclase grains are essentially unzoned.

A grid of laser ablation–inductively coupled plasma–mass spectrometer (LA-ICP-MS) analyses from garnet grain 4 in 12NZ04b indicates modest zoning in REEs and other trace elements (Fig. 6). Heavy (H) REEs and Y have high concentrations in the garnet core and lower concentrations toward the rims, forming broad bell-shaped zoning patterns. The outermost 0.5 mm of the garnet has higher Lu and flat Y concentration profiles. Garnet has low light (L) REE and high HREE concentrations throughout (Fig. 6E). However, unlike the REE in eclogite garnet, the core has flat REE patterns and the rim shows slight downward curvature in the HREE section of the chondrite-normalized HREE plots. Unlike eclogite garnet, this grain exhibits small negative to small positive Eu anomalies.

Pyroxene is fractured, subhedral to anhedral, and occurs as subrounded 1–2-mm-diameter grains. Compositional zoning is easily seen in thin section and X-ray maps as a change from low Ca and high Na in broad omphacitic cores to high Ca and Mg, and low Na content in narrow rims (Table 2). Similar to pyroxene in eclogite, the crystal cores commonly have abundant exsolution lamellae and rims lacking exsolution textures.

Amphibole occurs as masses of subequant grains with plagioclase, euhedral crystals in oriented growth with plagioclase, and as radiating prisms along grain boundaries of garnet and pyroxene. The amphibole is interpreted as late based on its association in plagioclase corona around garnet and pyroxene. A few early plagioclase grains can be found as anhedral inclusions in garnet and pyroxene. Late plagioclase is found in cracks within garnet where it is associated with plagioclase and biotite, and in grain boundary intergrowth with amphibole. Rutile is found as inclusions within major phases and in the matrix. Scapolite and clinozoisite are locally found in the matrix filling in spaces between pyroxenes and amphiboles.

Metadiorite Gneiss 12NZ09

Sample 12NZ09a is a plagioclase-rich gneiss collected from the southwest corner of Entry Island (Fig. 2). The sample is medium-grained, moderately foliated hornblende metadiorite gneiss. Matrix plagioclase displays prominent polygonal texture and is strongly saussuritized. Foliation is defined by aligned clots of hornblende, plagioclase, quartz, epidote, chlorite, and titanites, all of which are likely retrograde replacement products of garnet based on field observations from nearby garnet granulite gneiss. A network of thin, <1-mm-wide quartz veins crosscut the foliation.

Hornblende Gneiss 12NZ11

Sample 12NZ11a is medium- to fine-grained lineated gneiss of the Resolution Orthogneiss from the Gilbert Islands (Fig. 2). The rock is equigranular hornblende biotite gneiss with accessory titanite. There is no evidence for replacement of garnet or pyroxene; therefore, these minerals were never present or completely replaced with textural evidence destroyed.

Biotite Hornblende Gneiss 12NZ12

Sample 12NZ12b is a medium-grained gneiss from the eastern end of the Gilbert Islands (Fig. 2). The sample is medium-grained biotite hornblende metadiorite with a well-developed foliation defined by hornblende and biotite. The well-aligned grains display little intracrystalline strain suggestive of a hypersolidus foliation. Graded bedding and magmatic cross-bedding are prominent in outcrop. The sample contains trace amounts of apatite, zircon, rutile, and ilmenite, and secondary anhedral titanite after ilmenite and rutile. Similar to sample 12NZ11a, there is no evidence for replacement of garnet or pyroxene; we interpret that these minerals were never present in 12NZ12b and that the rock largely preserves igneous mineralogy.

Garnet Sm-Nd Isotope Data

One core and three rim aliquots were separated and hand-picked from large (∼1 cm) garnet in eclogite sample 12NZ01a. Clinopyroxene and fine-grained garnet surrounding the large garnet crystal were pulverized and then separated with a Frantz isodynamic magnetic separator. A single whole-rock fraction close to the large garnet was chosen to complete the isotope data set (Table 3). The garnet core, 2 garnet rims, 3 clinopyroxene splits, and the whole rock define a 143Nd/144Nd versus 147Sm/144Nd isochron with an age of 108.2 ± 1.8 Ma, mean square of weighted deviates (MSWD) = 1.6. A single garnet rim fraction with a 143Nd/144Nd value significantly higher than the ca. 108 Ma isochron was excluded from this calculation. This garnet rim combined with the three clinopyroxene fractions and the whole rock yields a much older age of 123.1 ± 2.8 Ma, MSWD = 1.6 (Fig. 7A).

Two cores were drilled and two rim fractions were obtained from two ∼1 cm garnet crystals in granulite sample 12NZ04b. Three additional garnet fractions obtained from a third irregular garnet grain (Grt 2–1, Grt 2–2, and Grt 2–3) in this sample. The irregular grain (Grt 2) could not be separated as rim and core fractions. A 143Nd/144Nd versus 147Sm/144Nd isochron based on all but one of the garnet fractions and the whole rocks define an age of 110.5 ± 1.6 Ma, MSWD = 1.4 (Fig. 7B). Similar to sample 12NZ01a, a single garnet fraction (Grt 2–3) has a significantly higher 143Nd/144Nd than the isochron and was excluded from the ca. 111 Ma age estimate. This garnet fraction and the whole rocks yield a substantially older age of 121.3 ± 2.9 Ma, MSWD = 0.0002 (Fig. 7A). Sm and Nd concentrations determined by isotope dilution–thermal ionization mass spectrometry (ID-TIMS) are in excellent agreement with concentrations determined by LA-ICP-MS. For example, ID-TIMS Nd concentrations vary from 1.7 to 2.8 ppm and LA-ICP-MS Nd concentrations vary from 1.5 to 3.8 ppm with the vast majority of analyses between 2 and 3 ppm (Fig. 7D).

Zircon Geochemistry and U-Pb Geochronology

Zircon in eclogite sample 12NZ01d are anhedral rounded with “soccer ball” morphology and range in size from 40 to 80 um. Cathodoluminescence (CL) images indicate weakly defined patchy and sector zoning with local blurred hints of oscillatory zoning. These zircon grains contain 2 to >200 ppm U (Table 4). The zircon spots yield mostly concordant 206Pb/238U ages that range from 67 to 1143 Ma with a lower intercept age of 95.5 ± 3.9 Ma (Fig. 8B). The 206Pb/238U ages produce an error-weighted average age of 95.2 ± 2.8 Ma (Fig. 8A). Individual spot ages of 497 and 1144 Ma are clearly xenocrystic and were not included in the weighted average age calculation. In addition, spot ages of 108.0 ± 7.3, 106.0 ± 2.9, and 67.0 ± 5.9 Ma are excluded from the weighted average age calculation because of discordance. The young discordant age (67 Ma) may reflect Pb loss and the two older results (497 and 1144 Ma) are interpreted to reflect inheritance from older metamorphic rocks, most likely the Takaka sequence.

Zircon in sample 12NZ04b are rounded and anhedral in shape and range in size from 200 to 150 um. They have complex irregular internal CL zoning dominated by patches of high and low CL response. The Tera-Wasserburg concordia diagram illustrates that ages are near-concordant from ca. 126 to 110 Ma, and form 2 clusters (Fig. 8E). Therefore, the data may represent a single age population with a weighted average of 117.5 ± 1.3 Ma (MSWD = 1.9) or 2 age populations estimated from lower concordia intercepts ca. 120 and 115 Ma. The Sambridge and Compston (1994) unmixing algorithm, assuming 2 populations, produces populations with ages of 119.0 ± 2.0 and 115.7 ± 2.0 Ma (Fig. 8F), indistinguishable from the 2 lower intercepts.

Zircon in sample 12NZ09b exhibit convoluted CL zoning, are anhedral to subhedral, and range in size from 100 to 200 um. Zircon grains yield 206Pb/238U dates ranging from 125.5 ± 1.8 to 109.0 ± 3.1 Ma. The Tera-Wasserburg concordia diagram illustrates that ages extend along concordia from ca. 126 to 108 Ma, forming clusters and intercepts ca. 123 and 111 Ma (Fig. 8E); therefore, a single weighted average is also inappropriate for this sample. Assuming 2 populations, the Sambridge and Compston (1994) unmixing algorithm produces populations with ages of 119.7 ± 1.4 and 111.4 ± 2.0 Ma (Fig. 8F) indistinguishable from the 2 lower intercepts.

Zircons from lineated quartz diorite gneiss from 12NZ11a exhibit oscillatory zoning with resorption textures. Most grains are prismatic, euhedral, and range in size from 100 to 200 um. The zircons have 206Pb/238U ages clustering on Tera-Wasserburg concordia at 93.6 ± 2.1 Ma (Fig. 8K). After excluding 1 discordant age, the remaining 20 206Pb/238U dates produce a weighted average age of 94.7 ± 2.0 Ma (Fig. 8L).

Zircons from hornblende biotite gneiss in 12NZ12b are subhedral and have subtle oscillatory zoning in CL (Fig. 8M). The zircons include a concordant cluster ca. 115 Ma and 4 younger dates extending to ca. 70 Ma. The 12 clustered dates give a weighted average 206Pb/238U age of 115.1 ± 1.6 Ma. The youngest 4 spot dates of 102.2 ± 1.6, 93.1 ± 3.0, 88.0 ± 2.1 and 71.7 ± 1.5 Ma are excluded from the weighted average age calculation due to slight discordance or young dates. Spots with dates of 93.1 and 88.0 Ma have low Th/U concentrations and may represent metamorphic zones in the crystals.

All of the analyzed zircon grains are depleted in LREEs and enriched in HREEs; however, there are distinct populations of zircon with anomalies in Ce and Eu (Fig. 9). Zircon in granulite from 12NZ4b and gabbroic gneiss (12NZ12b) show positive Ce and negative Eu anomalies. Zircons in intermediate gneiss (12NZ01d) collected adjacent to eclogite and quartz diorite gneiss 12NZ11a exhibit a positive Ce anomaly, but no negative Eu anomaly. Zircons from granulite on Entry Island sample 12NZ09b have two populations with distinctively different REE patterns. One pattern exhibits a positive Ce anomaly, negative Eu anomaly, and higher HREE concentrations. The second population exhibits an REE pattern with a positive Ce anomaly, a slightly positive Eu anomaly and much lower HREE concentrations. The second population is restricted to analyses from near the rims of zircon grains and has <0.78 Th/U ratios (Fig. 10). Zircons from granulite samples 12NZ01d and 12NZ11a exhibit low 232Th/238U and young 206Pb/238U ages. The Th/U ratios, Eu anomalies, and Yb concentrations indicate that zircon from samples 12NZ1d and 12NZ11a with the youngest zircons, 95–90 Ma, form a cluster distinct from the igneous population of zircons in the WFO (Fig. 10). Metamorphic zircons from metasedimentary rocks adjacent to the WFO and a few within the WFO range in age from 121 to 110 Ma (Gebauer, 2016; Schwartz et al., 2016) and form a broad field that extends from the 2 young samples to the WFO igneous field.

P-T Paths and Peak Metamorphic Conditions

Eclogite sample 12NZ01a was chosen for P-T-time forward modeling because garnet zoning indicates preservation of early garnet compositions and abundant clinopyroxene without amphibole is compatible with minimal retrograde reequilibration. A P-T pseudosection model was constructed from 600 to 1200 °C and 5–20 kbar (Fig. 11A) using the modified whole-rock composition in Table 5. The water content was chosen from a T versus mole fraction water phase diagram so as to reproduce the observed relatively dry mineral assemblages. The metamorphic assemblage (garnet + biotite + ilmenite + clinoamphibole + clinopyroxene + rutile + spinel) observed in the rock is predicted below the solidus (925 to ca. 1060 °C) at >10 kbar. The observed grossular, almandine, and pyrope garnet core isopleths do not intersect in the modeled P-T space. However, the compositional isopleths for the broad high grossular rim on garnet (Fig. 6A) intersect in a small area of the P-T pseudosection within the peak mineral assemblage field indicating metamorphic temperatures of 950 °C and ∼11 kbar.

Garnet granulite sample 12NZ04b was chosen for forward modeling because large subhedral garnet crystals may preserve early garnet compositions. However, in spite of the large crystals, no significant major element zoning was observed in garnet. A P-T pseudosection model was constructed from 600 to 1200 °C and 5–20 kbar (Fig. 11B) using the whole-rock composition in Table 5. The water content was chosen from a T versus mole fraction water phase diagram so as to reproduce the observed mineral assemblages. The peak metamorphic assemblage (garnet + clinopyroxene + biotite + plagioclase + rutile + melt) is predicted to be stable from ∼875 to 1075 °C and 14 to 19 kbar. However, the observed grossular, almandine, and pyrope garnet isopleths intersect at ∼1050 °C and 13–14 kbar, at the lower limit of the pressure range for the peak mineral assemblage field in the eclogite sample. The outcrop where this rock was collected contains leucosome suggesting partial melting compatible with the peak mineral assemblage prediction above the solidus.

DISCUSSION

Zircon and garnet ages combined with P-T estimates provide important new constraints on lower crustal processes recorded in the Breaksea Orthogneiss and adjacent rocks in Fiordland. These data are critically examined (see following) in order to distinguish between igneous crystallization and subsequent metamorphic growth events. The results underline the need for comprehensive isotope and trace element data sets for dating the complex metamorphic and igneous processes in high-temperature rocks and help to elucidate mechanisms for heating of lower crustal rocks.

Zircon 206Pb/238U Dates

The majority of zircon in 12NZ04b and 12NZ09b from the Breaksea Orthogneiss are anhedral with complex CL patterns and the U-Pb isotope data reflect this with multiple age populations in both samples. Tera-Wasserburg concordia plots clearly illustrate these complex isotopic signatures with dates that extend along concordia (overdispersed) that are inconsistent with single homogeneous populations (Fig. 8). Both samples have dates along concordia that produce two concordant intercept ages (Figs. 8D–8I). These results are interpreted as two discrete events with isotopic ratios between the intercepts reflecting mixed ages. The older ages of 123–120 Ma are interpreted as igneous ages from crystallization of the Breaksea Orthogneiss magma. All of these zircon lack HREE depletion (Fig. 9) and likely were not in equilibrium with garnet during growth. Younger zircon grains also lack HREE depletion; therefore, recrystallization in equilibrium with garnet is unlikely in spite of the presence of peritectic garnet. We interpret the younger 115–111 Ma 206Pb/238U ages to result from igneous grains that lost Pb during garnet granulite metamorphism.

Zircon in Resolution Orthogneiss sample 12NZ12b have subhedral crystal shapes and simple CL patterns with a tight cluster of U-Pb spot dates at 115 Ma, and 4 much younger spots that range to ca. 70 Ma. This sample preserves igneous layering and other textures suggesting minimum recrystallization and strain in spite of its location within the Resolution Island shear zone. Therefore, we interpret the 115 Ma age as igneous crystallization. This result is indistinguishable from the range of zircon ages, 118.0–115.4 (Schwartz et al., 2017), reported for the Malaspina pluton, and we speculate that the Resolution Orthogneiss may be a part of the Malaspina pluton.

Unlike the zircon described here, zircons in the garnet-free amphibole bearing samples 12NZ01d (Breaksea Orthogneiss) and 12NZ11a (Resolution Orthogneiss) contain subhedral zircon crystals with relatively simple age populations and single well-defined concordia intercept dates (Fig. 8). Sample 12NZ01d was obtained from an outcrop-scale high-strain zone that flanks the eclogite dome sampled nearby and sample 12NZ11 is from highly deformed rocks within the Resolution Island shear zone. Both samples are dominated by amphibolite facies minerals that are interpreted to reflect late-stage strain during diapiric flow in the lower crust (Betka and Klepeis, 2013). A few discordant spots may indicate older, likely inherited grains; however, data are insufficient for defining upper intercepts to concordia. We conclude that zircons in Breaksea Orthogneiss eclogite sample 12NZ01d and Resolution Orthogneiss sample 12NZ11a grew during late amphibolite facies metamorphism. Nearby samples of the Breaksea Orthogneiss on Hawea Island (samples 1333D and 1333E; Klepeis et al., 2016) have zircon age populations of 123.5 ± 1.4 and ca. 88.8 ± 4.7 Ma. The zircon rims from these Hawea samples yield a Ti-in-zircon temperature of 619 ± 60 °C (Klepeis et al., 2016). Therefore, these ages were interpreted to reflect initial pluton emplacement and later amphibolite facies conditions. The embayed zircon cores (ca. 123.5 Ma) are surrounded by distinct bright luminescent rims with low Th/U. The lack of evidence for older zircon and old embayed zircon cores in our Breaksea and Resolution samples is compatible with consumption of zircon prior to amphibolite facies metamorphism. Zircon resorption is likely to have occurred during the latter stages of granulite facies metamorphism. The late stage, ca. 90 Ma zircon rim growth is likely to result from fluid influx that led to hydrous amphibolite facies mineral growth.

Origin of Garnet Zoning

Metamorphic temperatures in most or all of the rocks around Breaksea Sound and throughout much of western and northern Fiordland exceeded 800 °C and locally exceeded 900 °C (De Paoli et al., 2009; Stowell et al., 2014) (Fig. 11). These temperatures are sufficient for significant intracrystalline diffusion of major and some trace elements in garnet (e.g., Ganguly et al., 1998a, 1998b; Carlson, 2006, 2012) and the WFO may have remained above 800 °C for ∼10 m.y. (Schwartz et al., 2016). However, the strongly zoned major and trace elements found in some of the Breaksea garnet crystals suggest that cooling rates were high enough that zoning was preserved. Cooling rates of ∼10 °C/m.y. estimated for the ∼700–800 °C part of the cooling path and higher cooling rates thereafter (Schwartz et al., 2016), and large garnet crystals (commonly 1 cm and locally >5 cm) likely precluded large-scale reequilibration of REE and other high field strength elements (e.g., Stowell et al., 2010, 2014).

The sharp boundary between low grossular–high pyrope cores and high grossular rims found in eclogite garnet (Fig. 6A) is interpreted to reflect partial recrystallization (without significant diffusion) of a high-temperature, possibly homogeneous, garnet grain because the high pyrope and almandine values in the core are compatible with high-temperature equilibration. The compositional zones in this eclogite garnet are somewhat similar to those described in omphacite granulite at Breaksea Tops (Type 1 in Clarke et al., 2013); however, the eclogite garnet described here has much wider (∼1.5 mm) rims with less almandine than the cores. This is unlike garnet described by Clarke et al. (2013) that has higher almandine in the rims. The eclogite garnet cores described here could be relict igneous or high-pressure eclogite facies garnet. Rims may be later metamorphic overgrowths or early garnet that changed composition by reequilibration during later granulite facies metamorphism. The similarity of major element concentrations in various textural types of garnet precludes definitive interpretation; however, eclogite cores described here are higher Fe and lower in pyrope and grossular than garnet interpreted as cumulate nearby in the Breaksea Tops area. The high-temperature eclogite garnet cores described here are interpreted to be relict compositions formed during eclogite facies metamorphism and preserved during the later granulite metamorphism that is responsible for the broad garnet rims.

The nearly complete lack of zoning of major elements in garnet from granulite sample 12NZ04b (Figs. 6B, 6C) is interpreted to have been produced by diffusion of major elements during or after peak metamorphic temperatures were reached. In this interpretation, no eclogite facies garnet compositions are preserved in sample 12NZ04b and the garnet major element compositions record late ca. 110 Ma metamorphism. Unlike major elements, some trace elements (e.g., Lu and Y) are strongly zoned (Figs. 6C, 6D), indicating preservation of trace element growth compositions in some garnet grains.

Garnet Growth and Trace Element Signatures

Migmatitic granulite and eclogite facies rocks present a significant challenge for interpreting their geologic histories (e.g., Clarke et al., 2013). This results from likely alteration of primary textures and compositions during the late stages of metamorphism, similar mineral compositions resulting from igneous crystallization and high-temperature metamorphism, and similar textures and mineral compositions resulting from partial melting during metamorphism and earlier igneous processes. Clarke et al. (2013) and Chapman et al. (2015) interpreted that much of the garnet from Breaksea Tops (Fig. 1) has an igneous origin. In their interpretation, garnet in clinopyroxenite and garnetite is igneous and the composition and zoning of this garnet reflects cumulate processes with local overgrowths on early cumulate grains. However, garnet in interlayered omphacite granulite and eclogite defines S1 “texturally classifying these grains as metamorphic” according to Clarke et al. (2013, p. 1934) who used major element compositions and chondrite-normalized REE patterns to interpret that the cores of omphacite granulite garnet retain igneous compositions, but that the rim compositions reflect later metamorphism. The smallest grains of garnet that form corona around omphacite were interpreted as entirely metamorphic in origin. The slope and curvature of HREE/chondrite in garnet and zircon may reflect equilibration between the two phases because both minerals strongly partition HREEs into the crystal and equilibrium between them results in competition for the HREEs (e.g., Rubatto, 2002). Breaksea granulite garnet in 12NZ04b has complex variations in HREE patterns with concave-downward curvature in HREEs for rims (Fig. 6G). This curvature in HREEs could reflect late growth in equilibrium with metamorphic zircon. The Eu anomalies in Breaksea garnet have been interpreted to indicate equilibrium with plagioclase (negative) and consumption of preexisting plagioclase (positive) to grow metamorphic garnet (Clarke et al., 2013). However, igneous crystallization of an oxidized magma could result in igneous garnet growth without a significant Eu anomaly (Trail et al., 2012). Crystallization of a less oxidized magma could produce garnet with a negative Eu anomaly because Eu2+ would be retained in the melt and not incorporated into crystallizing garnet. Metamorphic garnet growth in eclogite presumably would reflect the whole-rock composition and might inherit a Eu anomaly from a Eu-depleted igneous source. Garnet described here from eclogite and granulite in the Breaksea Orthogneiss has very small or no Eu anomalies, suggesting that either the oxidation state was high enough to result in a preponderance of Eu3+ and little or no Eu sequestered in plagioclase or a lack of plagioclase in equilibrium with garnet during garnet growth. In most eclogite, plagioclase only occurs locally with amphibole in reaction textures between clinopyroxene and garnet. However, plagioclase inclusions in garnet from granulite indicate that plagioclase was present during at least some garnet growth.

In summary, the major and trace element, and isotopic compositions of garnet described here do not allow a clear interpretation of igneous versus metamorphic origin for garnet. The whole-rock composition of eclogite from the Breaksea Orthogneiss (e.g., 12NZ01) is not a likely melt composition (Wiesenfeld, 2016) and must represent a cumulate. At ∼20 kbar, the pMELTS models for monzodiorite magma presented by Clarke et al., (2013) indicate that garnet would crystallize early followed by clinopyroxene. The predicted garnet compositions are similar to those described here and an igneous origin for some relict parts of garnet in the eclogite is possible.

P-T Conditions for Eclogite and Granulite Facies Metamorphism

The P-T conditions for eclogite facies are not well constrained in this study due to the pervasive mineralogical and compositional changes associated with the later granulite facies metamorphism. However, P-T pseudosections and garnet compositions indicate temperatures of 950–1050 °C at suprasolidus conditions for both eclogite and granulite. The mineral assemblage field for eclogite in sample 12NZ01a extend over a broad range of pressure, from 10 to >20 kbar and 600 to 950 °C. The peak mineral assemblage field for granulite in sample 12NZ04b extends from 13.5 to 19 kbar and from 850 to 1050 °C (Fig. 11). Quartz and rutile inclusions within garnet (Fig. 5B) and garnet compositions from 12NZ04b further constrain the P-T conditions. Quartz stability is predicted at and below the solidus above 14 kbar in this sample (garnet isopleth intersection in Fig. 11) and garnet compositional isopleths intersect at 12–14 kbar and ∼1050 °C. The simplest explanation for this is a clockwise P-T path from the peak mineral assemblage at peak pressure followed by possible exhumation and increasing temperature. This P-T path of increased temperature and slightly decreasing pressure is predicted to result in garnet growth. This path may reflect only the granulite facies event with little or no preservation of earlier eclogite facies mineralogy; however, the highest permissible pressure in the peak mineral assemblage field (with melt) is compatible with eclogite facies metamorphic conditions. Heating and reequilibration after exhumation from eclogite depths could have caused partial recrystallization of garnet. This would result in flat zoning profiles for major elements in the outer parts of garnet (Fig. 6A), the observed rutile inclusions in garnet (Fig. 5B), and the predicted ilmenite stability (including late ilmenite replacing rutile in the matrix) (Fig. 11). The 14–19 kbar pressures estimated here overlap with the 15.5–19 kbar pressures estimated by De Paoli et al. (2009) and are compatible with local preservation of minerals that equilibrated at peak pressure and partial reequilibration during granulite and later metamorphism.

Uncertainties in P-T estimates preclude definitive evaluation of whether the eclogite and granulite west of the Resolution and Straight River shear zones equilibrated at the same pressure, because the eclogite pressure estimates range from 14 to 19 kbar. However, the early high-pressure estimates for granulite presented here are compatible with the conclusion of De Paoli et al. (2009) that both lithologies underwent high-pressure eclogite facies metamorphism at ∼18 kbar. The more precise estimate of 12.5–15 kbar for later granulite facies metamorphism is indistinguishable from pressure estimates for garnet granulite in the Malaspina pluton. This indicates that all of the WFO rocks, across the Resolution Island and Straight River shear zones, from Breaksea Sound north to Doubtful Sound (and likely further), were affected by similar granulite facies conditions. This requires that any large vertical displacement on the two shear zones must have been prior to ca. 110 Ma. Pressure estimates from this study are similar to those of De Paoli et al. (2009), who estimated 720–800 °C at 15.5–18 kbar for nearby eclogite and granulite. However, our estimates for the later granulite facies event are 100–200 °C higher in temperature and range to ∼3 kbar lower pressures.

Garnet Ages and Diffusion in Garnet

REE-bearing inclusions identified in garnet from eclogite and granulite include apatite, rutile, and clinopyroxene. In addition, sparse zircon grains have been identified in the rock matrix; therefore, we cannot rule out some contamination of garnet with some of these mineral inclusions. Hand-picking to remove garnet with obvious inclusions followed by aggressive HF leaching is interpreted to have removed most of the inclusions. This is based on close similarity of the Sm and Nd concentrations and the Sm/Nd ratios obtained from isotope dilution and LA-ICP-MS (Data Repository Table DR1).

Garnet from the Breaksea Sound area underwent temperatures in excess of 800 °C; therefore, diffusion is likely to have partially or wholly changed some elemental concentrations and ratios after crystal growth. The significance of REE diffusion has been evaluated by comparison of observed trace element zoning in garnet to that observed elsewhere, with simple diffusion models, and by estimating closure temperatures. Although REE diffusion mechanisms and rates in garnet may vary considerably, data from Van Orman et al. (2002), Carlson (2012), Smit et al. (2013), and Bloch et al. (2015) are in good agreement. These studies indicate that all of the REE diffusion coefficients are 1–1.5 log10 units less than those for the divalent cations (Carlson, 2012). Therefore, major element zoning, which may reflect diffusion, is likely to be decoupled from that of the REEs, which may reflect growth. Trace element zoning in granulite garnet (12NZ04b; Fig. 5B), which has a distinctive zone of low Y near the rim and insignificant zoning in major elements, is compatible with decoupling of REE and major elements. The low-Y rim in granulite garnet is reverse to that described in Carlson (2012), who interpreted high-Y rims as reflecting diffusive incorporation of Y into garnet during resorption at high temperature, because no minerals in the rock are likely to fractionate Y more strongly than garnet (e.g., Johnson, 1998). Therefore, we tentatively interpret the observed Y zoning in Breaksea granulite to reflect growth without Y diffusion. In this case, significant REE diffusion is unlikely because REE diffusion coefficients are very similar to Y (Carlson, 2012). The possible effects of diffusional resetting can also be evaluated in terms of closure temperature (Dodson, 1973), which yields useful approximations for REE diffusion at cooling rates of ∼10 °C/m.y. and garnet radii of <0.2 cm (Ganguly et al., 1998b). Using the lowest permissible cooling rate of ∼10 °C/m.y. (Schwartz et al., 2016) and effective radii of garnet diffusion ranging from 0.1 to 0.5 cm (the smallest grains used for geochronology), closure temperatures (Tc) for Nd are 900–1030 °C (see Data Repository Table DR2 for details). Note that Tc for the larger grains may be overestimated due to departure from the assumptions required for quantification of Tc (see Ganguly et al., 1998b). Regardless, the observations are compatible with a lack of diffusive reequilibration resulting from the large garnet grains and relatively rapid cooling from high temperature (e.g., Schwartz et al., 2016).

Only one garnet aliquot of each dated sample preserves isotopic compositions synchronous with the ca. 122 Ma zircon ages that are interpreted as igneous. We are unable to document the spatial extent of these old regions in garnet and do not have an exact explanation for why the older age aliquots were obtained near the garnet rims. However, the grains are locally irregular in shape and the old aliquots may not be from grain edges. Assuming that eclogite and garnet have the same temperature history, with eclogite metamorphism at ∼800 °C and granulite at ∼1000 °C, then the rocks heated at 15 °C/m.y. The high grossular rims on eclogite garnet grew at the granulite facies conditions and subsequent cooling must have been rapid in order to preserve these rims.

Granulite Facies Metamorphism, Garnet Growth, and Lower Crustal Evolution

Garnet from eclogite and granulite in the Breaksea Sound area have 2 age populations: 2 garnet fractions that have 123–121 Ma ages, and a larger number of garnet fractions have younger 111–108 Ma ages (Table 3). The two older ages could reflect igneous crystallization from the parental diorite to monzonite magma or the early eclogite metamorphic event. These ages are indistinguishable from the oldest 206Pb/238U zircon ages reported for the Breaksea Orthogneiss in Klepeis et al. (2016) and herein. Garnet growth ages of 126–123 are reported for the Pembroke Granulite ∼150 km north of Breaksea (Stowell et al., 2010); however, no other metamorphic ages in this range have been reported south of Milford Sound. Synchroneity of the older garnet ages for Breaksea Orthogneiss and the oldest 206Pb/238U zircon ages for the Breaksea Sound area (Table 6), which we interpret as igneous, is compatible with interpretation of the old garnet as igneous. This would be in line with the interpretation of Clarke et al. (2013) that some of the garnet from Breaksea Tops is igneous. However, eclogite facies metamorphic garnet growth ca. 123–121 Ma is also possible and we consider this possibility here. Metamorphic textures suggest that the majority of garnet is metamorphic and we interpret that the 111–108 Ma garnet Sm-Nd ages date garnet granulite metamorphism.

Data presented here for garnet and zircon provide important constraints on lower crustal architecture and history. Thermobarometry (De Paoli et al., 2009) and P-T pseudosections indicate that at least some of the garnet in Breaksea rocks preserves evidence for equilibration at ∼18 kbar. As discussed herein, garnet trace element compositions and other data do not provide unambiguous evidence for distinguishing igneous versus metamorphic garnet growth. However, garnet rims must have been equilibrated at temperatures >850 °C ca. 111 Ma based on the rim garnet compositions presented here (Fig. 11B) and Sm-Nd ages (Fig. 7). This leads to two plausible scenarios: (1) high-pressure garnet cumulate formation ca. 122 Ma, ∼14 kbar granulite facies metamorphism ca. 110 Ma, cooling and/or amphibolite facies metamorphism at 95–90 Ma; and (2) cumulate formation at or before 122 Ma, 18 kbar eclogite facies metamorphism ca. 122 Ma, ∼14 kbar granulite metamorphism at 110 Ma, and cooling and/or amphibolite facies metamorphism at 95–90 Ma. In both of these scenarios, the amount and timing of cooling between granulite and amphibolite facies metamorphism is uncertain and water must have been added to the rocks sometime after the peak of granulite facies metamorphism.

Documenting igneous versus metamorphic garnet could have significant impact on interpretation of metamorphism and the thermal structure of the crust. Cumulate garnet growth is most compatible with one or more layered magma chambers and a hot lower crust dominated by suprasolidus processes, which would have cooled to temperatures required for garnet crystallization. Metamorphic garnet growth could occur at the same temperatures as igneous crystallization, with significant growth at subsolidus temperatures. Therefore, the lowermost crust would likely have been significantly hotter at 122 Ma if the garnet is suprasolidus. Large garnets associated with leucosomes in granulite components of the Breaksea Orthogneiss are clearly porphyroblastic (Figs. 3 and 4) in texture and are not dominantly associated with minerals that would crystallize as cumulate phases in mafic rocks (e.g., Type 3 garnet in Clarke et al., 2013). Yet these grains preserve old components which are the same age as the oldest eclogite garnet, indicating that the oldest parts of garnet in both eclogite and granulite are likely of metamorphic origin.

The 111–108 Ma metamorphic garnet is similar to but younger than the majority of garnet dated in the Malaspina pluton to the north and east of Breaksea Sound. Comparison of the average age of garnet in the Malaspina pluton with the Breaksea metamorphic garnet growth ages (Fig. 12) indicates that all of the WFO, across the Resolution Island and Straight River shear zones, from Breaksea Sound north to Doubtful Sound (and likely further), were affected by the same granulite facies metamorphic event, and tentatively that garnet was growing or reequilibrating in the Breaksea rocks 1–2 m.y. later than in the Malaspina pluton. The two ca. 95 Ma amphibolite facies zircon ages are in agreement with high temperatures remaining in the Breaksea Sound rocks west of the Resolution Island shear zone, while rocks to the east may have cooled. Zircon and titanite from two ductilely deformed calcsilicate rocks in the Resolution Island shear zone separating the Breaksea and Resolution Orthogneisses from the Malaspina have been dated (Fig. 2; Schwartz et al., 2016). The 206Pb/238U ages for both of these samples are ca. 113 and 110 Ma, for zircon and titanite, respectively. Zirconium-in-titanite temperatures of 850–900 °C (Schwartz et al., 2016) and the titanite ages indicate high temperatures toward the end of garnet growth in the Malaspina pluton, and during garnet growth in the Breaksea Orthogneiss to the west.

Assuming that the oldest Breaksea garnet ages are metamorphic, there is no evidence that ca. 122 Ma high-temperature eclogite facies metamorphism at 18 ∼kbar affected Takaka metamorphic rocks surrounding the WFO. Metasedimentary rocks in the Deep Cove Gneiss of the Takaka Group contain abundant evidence for preservation of pre-Cretaceous metamorphic minerals (Allibone et al., 2009b; Chavez et al., 2007) that are unlikely to have survived protracted high temperatures. Later granulite facies metamorphism ∼14 kbar was pervasive in the lower crust at ca. 112 Ma (e.g., Stowell et al., 2014; Schwartz et al., 2016). Similarity of the younger garnet ages in the Breaksea Orthogneiss with those in the Malaspina pluton indicate that this high-temperature granulite event affected >600 km2 of the lower arc crust. Sparse garnet ages from the Misty pluton (Stowell et al., 2013; Yelverton et al., 2015) allow synchronicity of granulite facies metamorphism across an additional 670 km2. Metamorphic zircon rims from the Worsely pluton indicates a similar age for metamorphism of the Worsley pluton (Gebauer, 2016; Schwartz et al., 2017). Together, these data indicate that ∼1900 km2 of the lower crust was at 800–900 °C simultaneously. This large area of lower crust at granulite facies temperatures of >800 °C has important implications for crustal strength and continuity of the currently exposed crustal section. The extensive layer of high-temperature rock likely resulted in low strength and viscosity, allowing subsequent extensional collapse. Extension and lower crustal thinning was accommodated by ductile shear in numerous high-strain zones involving combinations of lateral and vertical flow (e.g., Klepeis et al., 2016). The lack of extensive ca. 112 Ma metamorphic signatures compatible with limited heating of the Deep Cove Gneiss, which is juxtaposed with the Malaspina pluton and Breaksea Orthogneiss (Chavez et al., 2007; Allibone et al., 2009b), requires considerable loss of structural section above the WFO or rapid cooling of the entire lower crust. If vertical displacements on the Straight River and Resolution Island shear zones were pre–110 Ma, then this loss of structural section must have been accommodated on other shear zones. The Doubtful Sound shear zone remains a viable candidate.

The Breaksea rocks must have been at or above 800 °C from the time of early eclogite garnet growth at 122 Ma until the growth of granulite garnet at 110 Ma, were exhumed from ∼65–45 km depths, and were at 620–700 °C during amphibolite facies metamorphism at 95–90 Ma. The temperature estimates for 18 kbar eclogite and 14 kbar granulite facies metamorphism combined with garnet ages for metamorphism are compatible with an ∼15-km-thick lower crustal section that was >800 °C. Kilometer-scale gneiss domes within the Breaksea Orthogneiss are locally cored by interlayered granulite and eclogite (Betka and Klepeis, 2013) and flanked by retrograde amphibolite facies gneiss. We speculate that diapiric rise within these domes juxtaposed eclogite and granulite facies mineral assemblages, which formed ca. 122 and 110 Ma, respectively. In this model, high-pressure eclogite facies rocks crystallized at 65 km ca. 122 Ma, then flowed upward nearly isothermally to ∼50 km, where mineral assemblages partly reequilibrated at 14 kbar ca. 110 Ma. Monzodiorite gneiss (e.g., 12N04b) may have flowed upward with eclogite and then recrystallized at lower pressure (e.g., De Paoli et al., 2009), or these rocks may be metamorphosed plutons that were emplaced at 50 km and then juxtaposed with eclogite during diapir emplacement. The later interpretation contrasts with that of De Paoli et al. (2009), who described high-pressure granulite and interpreted that both lithologies were metamorphosed at high-pressure eclogite facies conditions followed by lower pressure granulite facies metamorphism.

Results from this study, summarized in Figure 12, allow calculation of exhumation and heating rates. We use ∼18 kbar pressure for eclogite facies metamorphism, 11 kbar for granulite facies metamorphism, and garnet ages of 122 and 110 Ma. These data result in an exhumation rate of 2.2 km/m.y. Exhumation rates after granulite facies metamorphism cannot be estimated with the current data because robust pressure estimates for amphibolite facies metamorphism are lacking. The differing peak temperature estimates for eclogite and granulite facies metamorphism require heating of ∼100–190 °C during exhumation pre–110 Ma, thus, a heating rate of 4–16 °C/m.y. between 122 and 110 Ma. A simple cooling path from granulite to amphibolite facies temperatures of ∼650 °C require rapid, ∼23 °C/m.y. cooling from 110 to 95 Ma. This is near the high end of the 7–21 °C/m.y. cooling rate estimated for the Malaspina and Misty plutons (Schwartz et al., 2016) to the north and east of Breaksea Sound. A more complex path with cooling below amphibolite facies temperatures and reheating is permissible, but would require high rates of cooling and reheating.

The ca. 110 Ma high-temperature granulite facies metamorphism documented for eclogite and granulite in the Breaksea Orthogneiss is synchronous with or slightly younger than ca. 112.5 Ma granulite metamorphism throughout much of the lower arc crust of Fiordland. This event postdates the voluminous arc magmatism in the Misty and Malaspina plutons (Schwartz et al., 2017), requiring that metamorphism was driven by a heat source younger than these 118–115 Ma intrusions. Plutons of this age are not common anywhere in Fiordland and renewed heat flow from the mantle heating is a possible cause for the granulite facies event. Similar high-temperature metamorphism postdates voluminous pluton emplacement at mid- to lower crustal depths in arc systems throughout the world (e.g., Kenah and Hollister, 1983; Jagoutz and Schmidt, 2012) and throughout much of Earth history (e.g., Wang et al., 2011; Zhang et al., 2013; Liu et al., 2016). Depine et al. (2008) presented a thermal model indicating that continental arc crust could be near isothermal (∼800 °C) from ∼10 km depths to the Moho. If the entire mid- to lower crustal column of the Fiordland arc was at or above 800 °C ca. 120 Ma, then it could have remained hot until 110 Ma without significant new heat input. Younger, ca. 95 Ma, magmatism (Tulloch et al., 2009) indicates the possibility of the crust remaining hot significantly longer. The magmatic pulse that resulted in emplacement of the Misty and Malaspina plutons and high mantle heat flow could have been triggered by slab tear, ridge-trench collision, and/or delamination of a dense root. We conclude that the metamorphic history presented here is compatible with postmagmatic high heat flow resulting from slab tear or ridge-trench collision (Schwartz et al., 2017).

CONCLUSION

Geochronologic and thermobarometric data from the Breaksea Orthogneiss record the timing of eclogite, granulite, and amphibolite facies metamorphism in the lower crust of the Fiordland magmatic arc. Sm-Nd garnet and U/Pb zircon ages document early eclogite facies metamorphism ca. 120 Ma, followed by garnet granulite metamorphism between 111 and 108 Ma. The latter event was widespread and affected >600 km2 of the lower crust in the Cretaceous magmatic arc. Later ca. 95 Ma amphibolite facies metamorphism caused local hydration and replacement of garnet and pyroxene by biotite and amphibole. The new ages combined with pressure-temperature-time paths indicate magma intrusion into the lowermost arc crust, near isothermal exhumation of Breaksea rocks at ∼2.2 km/m.y. from ∼65 km to 40–45 km depths, followed by continued high heat flow and granulite facies metamorphism. We conclude that Sm-Nd garnet geochronology can provide useful ages for high-temperature rocks when large grains cool at rates of >10°C/m.y. However, the complex U-Pb zircon and Sm-Nd garnet data underline the need for comprehensive data sets from multiple rocks for deciphering the intrusive and subsequent thermal history of the lower crust. The geochronological data indicate that voluminous magmatism in Fiordland was closely followed by high-temperature metamorphism, a common phenomenon in the lower crust of magmatic arcs.

ACKNOWLEDGMENTS

This work would not have been possible without financial support from U.S. National Science Foundation grants EAR-1119039 (Stowell and Schwartz) and EAR-1352021 (Schwartz) and logistical help from GNS Science, New Zealand. Additional funds were provided by the University of Alabama Department of Geological Sciences Advisory Board, Hooks Fund, and Graduate School Research and Travel Support Fund. Drew Coleman (University of North Carolina at Chapel Hill) and Elizabeth Bollen and Karen Odom Parker (University of Alabama) provided assistance with Sm-Nd isotope data collection. Joe Wooden and Matt Coble provided assistance with U-Pb isotope data collection on the SHRIMP-RG (sensitive high-resolution ion microprobe–reverse geometry) at Stanford University. The New Zealand Department of Conservation TeAnau office graciously permitted access and sampling in Fiordland National Park. Richard and Mandy Abernathy provided invaluable assistance with Fiordland sampling. We thank Kurt Stuewe for helpful comments and efficient editorial handling. Matthijs Smit and Mike Williams provided constructive and helpful reviews. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

1GSA Data Repository Item 2017338, supplemental information on methods and tables containing REE concentrations and garnet closure temperature estimates, is available at http://www.geosociety.org/datarepository/2017, or on request from editing@geosociety.org.