Accessory mineral petrochronology reveals 30 m . y . of partial melting during the separation of Zealandia from eastern Gondwana

Granitic pegmatites emplaced within the Alpine Schist of New Zealand provide an opportunity to assess the relationship between metamorphism and melting during the separation of Zealandia from eastern Gondwana. We combined monazite, xenotime, and zircon petrochronology from 12 pegmatite samples with whole-rock geochemistry to link pegmatites to the conditions and mechanisms of partial melting of the Alpine Schist. Monazite dates define a broad range from ca. 85 to ca. 50 Ma, interpreted to represent primary crystallization, remelting of existing pegmatites, and fluid-assisted recrystallization. Xenotime and zircon dates match monazite dates from the same sample, except where recrystallization or sampling of inherited age domains has led to younger or older dates, respectively. The total age range of primary, unmodified igneous monazite is 79.5 ± 0.4–49.8 ± 0.2 Ma, extending the known lower time limit for pegmatite emplacement by 17 m.y., and indicating that the generation and crystallization of melts occurred over a period of 30 m.y. Whole-rock compositions indicate that water-fluxed melting was the dominant melting mechanism, although minor dehydration melting may have contributed to the earliest melts. Partial melting initiated immediately prior to the youngest record of garnet growth in the Mataketake Range region and persisted for 28 m.y. after garnet growth ceased in this region. The timing, duration, and mechanisms of partial melting and their relationship to the timing of metamorphism suggest that short-lived melting events were driven by episodic water fluxing from ca. 80 to 50 Ma in the Alpine Schist. Late Cretaceous to Paleogene intraplate deformation focused between the NW and SE regions of Zealandia is proposed as a potential mechanism for this prolonged period of partial melting and melt emplacement within a geographically restricted area. LITHOSPHERE; v. 11; no. 1; p. 169–189; GSA Data Repository Item 2019030 | Published online 20 December 2018 https://doi.org/10.1130/L1021.1


INTRODUCTION
The Alpine Schist of the South Island, New Zealand, represents some of the youngest material to be accreted at the Pacific-Gondwana subduction margin during Cretaceous orogenesis (Cooper and Palin, 2018).The metamorphic record preserved in the Alpine Schist postdates the widely accepted estimate for the end of subduction along this section of the Gondwana margin (ca.105 Ma), suggesting that metamorphism was synchronous with regional extension in Zealandia and the final stages of the breakup of Gondwana (Mortimer and Cooper, 2004;Vry et al., 2004;Cooper andIreland, 2013, 2015;Scott et al., 2015;Briggs et al., 2018).Peraluminous granitic pegmatite dikes and sills exposed in an ~5 km by ~27 km region of the Alpine Schist in the Mataketake Range region represent the only evidence of anatectic melting of the Alpine Schist (Wallace, 1974;Batt et al., 1999;Chamberlain et al., 1995).Limited existing geochronology data suggest that the pegmatites range from 82 to 67 Ma (Chamberlain et al., 1995;Batt et al., 1999), postdating prograde garnet growth in this region by up to 15 m.y.(Vry et al., 2004;Briggs et al., 2018).However, it is unclear whether this age range represents multiple discrete pulses of melt, semicontinuous melt production, or Pb mobility by diffusion or recrystallization in the dated accessory minerals.The young ages of metamorphism and partial melting of the Alpine Schist are enigmatic and difficult to explain based on the existing framework for the Cretaceous tectonic history of Zealandia, justifying a reassessment of the drivers of metamorphism and melting in this region.
To better determine the true tempo of melt production and further investigate the relationship between melting and metamorphism in the Alpine Schist, we dated accessory minerals from 12 pegmatites sampled over the ~5 km by ~27 km area, representing the entire geographic and geochemical extent of observed pegmatite occurrences.Detailed petrographic analysis was combined with U-Th/Pb petrochronology of monazite, xenotime, and zircon via laser-ablation split-stream inductively coupled plasma-mass spectrometry (LASS-ICP-MS), and wholerock geochemistry in order to link pegmatites to both the conditions and mechanisms of melting.The results presented here constrain the timing, duration, and mechanisms of partial melting in the Alpine Schist, and they provide insights into the geodynamic scenarios under which such protracted partial melting is possible.

GEOLOGICAL BACKGROUND
The Alpine Schist is a steeply dipping belt of greenschist-to amphibolite-facies schists and gneisses, exposed along the eastern side of the RESEARCH Alpine fault in the Southern Alps of New Zealand (Fig. 1).It is part of an assemblage of terranes that were accreted along the long-lived Paleozoic-Mesozoic eastern Gondwana margin.These terranes represent the basic components of a subduction margin, including the continental arc, forearc basin, and accretionary wedge complex, and they now comprise the basement rocks of the South Island of New Zealand (Mortimer, 2004).Detrital zircon ages from the Alpine Schist are as young as ca.108 Ma, indicating that deposition of the protolith accretionary wedge sediments immediately preceded the widely accepted estimate for the end of subduction along this section of the Gondwana margin (ca.105 Ma); these young protoliths are interpreted to represent the exotic Pounamu terrane (Cooper andIreland, 2013, 2015;Cooper and Palin, 2018).High-temperature geochronology from the Alpine Schist has produced a broad range of ages for the timing of metamorphism.Lu-Hf garnet ages range from 97.3 ± 0.3 Ma to 75.4 ± 1.3 Ma (Vry et al., 2004;Briggs et al., 2018), and a Sm-Nd isochron age of 100 ± 12 Ma was reported from the southern Alpine Schist (Mortimer and Cooper, 2004).Two monazite dates of 82.9 ± 2.3 Ma and 82.6 ± 1.0 Ma from Fox Glacier are in agreement with a nearby garnet age of 79.9 ± 2.3 Ma from Hare Mare Creek (Scott et al., 2015;Briggs et al., 2018).Zircon rims from the Pounamu ultramafic belt and Mataketake Range region yield younger ages of 72-64 Ma (Mortimer and Cooper, 2004;Cooper andIreland, 2013, 2015;Cooper and Palin, 2018).As these zircon rim dates are not yet reliably linked to any metamorphic stage, it is unclear whether they record high-temperature metamorphic growth or low-temperature hydrothermal (re)crystallization.
The exposed Alpine Schist has been uplifted along the active Alpine fault plate boundary within the past ~5 m.y.(e.g., Batt et al., 1999), revealing a tilted section (Fig. 1B) through metamorphic isograds.The recent and ongoing uplift of deep crust along the Alpine fault has resulted in a 1-km-wide mylonite overprint on the Alpine Schist east of the Alpine fault, which reinforced the dominant foliation that was formed during Mesozoic metamorphism (Little et al., 2002;Norris and Cooper, 2003).In the Mataketake Range and Mount Kinnaird area (Fig. 1), the Alpine Schist of the Pounamu terrane consists of amphibolite-facies schists and gneisses that are intruded by variably deformed granitic pegmatite dikes and sills (Fig. 2; Wallace, 1974).Within the context of local Alpine Schist isograds, the Mataketake Range falls within the oligoclase amphibolite and K-feldspar amphibolite metamorphic zones, which are equivalent to the Barrovian sillimanite zone.The dominant foliation of the Alpine Schist trends NNE (025°), oblique to the NE-trending Alpine fault.Existing pressure-temperature constraints for the Alpine Schist in the Mataketake Range region are 4.7-7.3kbar at 625-680 °C for K-feldspar zone schist (Wallace, 1974;Grapes, 1995;Beyssac et al., 2016), and 6.1 ± 2 kbar at 558 ± 50 °C for lower-grade (oligoclase zone) schist from the Moeraki River (Scott et al., 2015).The similarity between the initial 87 Sr/ 86 Sr values of the pegmatites and the nearby Otago Schist (0.7053-0.7060 and 0.7061-0.7098,respectively) supports the interpretation that pegmatites were derived from partial melting of the Alpine Schist country rock, since mantle xenoliths show that the underlying mantle is much less radiogenic in Sr (Wallace, 1974;Aronson, 1965;Chamberlain et al., 1995;Batt et al., 1999;Scott et al., 2014).Subsequent studies have made the assumption that this partial melting was concomitant with, or immediately followed, the attainment of peak metamorphic temperatures (e.g., Batt et al., 1999;Mortimer and Cooper, 2004;Scott et al., 2015).Existing zircon and monazite ages from four pegmatite samples obtained by U-Pb sensitive high-resolution ion microprobe analysis (SHRIMP) and isotope dilution-thermal ionization mass spectrometry (ID-TIMS) showed significant scatter both within and between samples, with reported ages ranging from 82 to 67 Ma (Chamberlain et al., 1995;Batt et al., 1999).

FIELD RELATIONSHIPS, PETROGRAPHY, AND MINERALOGY
The region containing pegmatite dikes and sills is located ~3 km southeast of the NE-SW-trending Alpine fault.The zone where pegmatites are observed is ~5 km wide and extends for at least 27 km along strike parallel to the Alpine fault, from the central Mataketake Range in the south to Doughboy Creek in the north (Fig. 1).The highest observed density of pegmatites occurs in the Mount Kinnaird region, where 0.5-12-m-thick sills comprise ~50% of a 0.5 × 0.5 km section of outcrop.Pegmatite exposures decrease to the southwest, where outcrop is limited.At their northwesternmost extent, pegmatites strike into the NE-trending Alpine fault mylonite zone, where they become progressively deformed and attenuated (Norris and Cooper, 2003).The limited extent of the pegmatites in the Alpine Schist has been suggested to reflect a depth-transgressive uplift profile of the Alpine Schist along strike of the Alpine fault, exposing a greater thickness of the metamorphic pile in the south than in the north (Grapes andWatanabe, 1992, 1994;Scott et al., 2015).
The majority of pegmatites occur as 0.15-30-m-thick sills conformable to the dominant foliation of the Alpine Schist (Fig. 2A), although rare dikes do crosscut sills at some locations and are isoclinally folded (Fig. 2B).Sills can be traced for up to 30 m along strike in the Mount Kinnaird region, where they form interconnected networks with sill width decreasing along strike to the north and south (Fig. 2A).Quartz veins <10 cm thick occur within some pegmatite sills, oriented parallel to sill margins (Fig. 2C).Pegmatite margins are gradational, with irregular, patchy boundaries (Fig. 2D).Sill margins pinch and swell and are often boudinaged (Figs. 2E and 2F).Leucosomes <2 cm thick occur within the gneissic host rock (Fig. 2G).A weak fabric is present in all pegmatites, defined by the alignment of micas oriented parallel to the dominant foliation in the host schist (Fig. 2D).Two of the 12 pegmatite samples, DC2 and MR13-3, were collected from within the Alpine fault mylonite zone (<1 km from the Alpine fault; Norris and Cooper, 2003).Deformation in these pegmatites is defined by extensive subgrain formation in quartz, grain-size reduction concentrated in the rims of feldspar augen, and large mica fish, giving these rocks the appearance of protomylonite (Figs. 3A and 3B).
Mineralogically, pegmatites consist of medium-to coarse-grained quartz, plagioclase, K-feldspar, and varying proportions of muscovite, biotite, and garnet, with accessory tourmaline, apatite, zircon, monazite, and xenotime (Fig. 3).Within 1 cm of a pegmatite margin, the host schist contains muscovite and biotite that lack reaction textures in association with quartz, plagioclase, and K-feldspar, which are recrystallized and show cuspate morphologies (Fig. 3C).Directly adjacent to this pegmatite margin, muscovite shows reaction textures along with quartz-plagioclase symplectites (Fig. 3D).Distinct melanosomes were not observed at pegmatite margins; however, in several localities, increased modal proportions of biotite and garnet were observed in the wall-rock schist within 5 cm of the sill margins.

ANALYTICAL METHODS
Twelve pegmatite samples (Table 1) were collected throughout the pegmatite zone, to increase the chance of identifying any spatial trends in age or geochemistry of pegmatites.Twelve associated quartzofeldspathic and amphibolite schist samples were also collected.One of these schist samples, MR13-30 (Table 1), contained sufficient accessory monazite to be analyzed.

RESEARCH
Crushed samples were powdered and cast into lithium metaborate fused glass disks.Major and minor oxides were analyzed by inductively coupled plasma-atomic emission spectrometry (ICP-AES; ALS Minerals procedure ME-ICP06), while trace elements were analyzed by solution ICP-MS (ALS Minerals procedure ME-MS81).

Laser-Ablation Split-Stream (LASS) ICP-MS
Zircon, monazite, and xenotime separates from 12 >2 kg pegmatite samples were prepared using standard techniques of crushing, panning, and heavy liquid and magnetic separation.Monazite in one schist sample (MR13-30) was analyzed in thin section.At least 30 crystals of monazite ranging from 50 to 1000 µm in diameter were handpicked from each sample and sorted into two size fractions; these were mounted in two epoxy resin grain mounts and polished to expose the inner portion of each crystal.Five samples contained significant proportions of green bipyramidal xenotime, of which three samples were analyzed (MR13-5, MR13-12, MR13-14).Only four samples contained sufficient zircon for dating (DC2, MR13-14, MR13-20, MR13-27), which included ~100-µm-long colorless prismatic crystals, and larger, stout, opaque crystals.
Polished equatorial sections of accessory minerals were imaged prior to analysis to reveal compositional zoning and guide U-Th/Pb spot locations.X-ray element maps of monazite and xenotime crystals were created using a Cameca SX-100 electron probe micro-analyzer (EPMA) at the University of California-Santa Barbara (UCSB).Element maps of Th, La, Pr, U, and Y for monazite and Ce, Th, Dy, La, and U for xenotime were produced using an accelerating voltage of 20 kV, beam current of 200 nA, and step size of either 1 or 2 µm.Zircon was mapped using a cathodoluminescence (CL) detector on a FEI Q400 FEG scanning electron microscope (SEM) at UCSB using methods outlined by Cottle (2014).
Accessory minerals were analyzed by LASS-ICP-MS at UCSB (Cottle et al., 2013;Kylander-Clark et al., 2013;with modifications outlined in McKinney et al., 2015).Samples were ablated with a Photon Machines 193 nm excimer laser system, after which the aerosol was split between a Nu Plasma high-resolution (HR) multicollector (MC) ICP-MS to measure U-Th/Pb, and an Agilent 7700S quadrupole ICP-MS (monazite and xenotime) or a Nu Instruments AttoM single-collector ICP-MS (zircon) to measure trace elements (rare earth elements [REEs], Si, P, Ca, Sr, Y, Hf).Approximately 107 monazite crystals were analyzed, with around five measurements made on each crystal to obtain ~40-50 measurements per sample.Six samples were analyzed in one session, and the remaining five analyses were carried out in a later session, using a 9.7 µm spot diameter, a frequency of 4 Hz, a laser energy of 3 mJ, and 25 s ablation time.Samples with two monazite size fractions were analyzed in the same session.Monazite and xenotime unknowns were bracketed by three natural reference monazite standards: 44069 (424.9 ± 0.4 Ma 206 Pb/ 238 U age; Aleinikoff et al., 2006) was used as a primary reference to correct for mass bias, downhole fractionation, and instrumental drift; FC-1 (55.7 ± 0.7 Ma 206 Pb/ 238 U ID-TIMS age; Horstwood et al., 2003) was treated as a secondary reference in every run as an independent assessment of the reproducibility and accuracy of the corrections; and Bananeira (508.9 ± 0.9 Ma 206 Pb/ 238 U LA-ICP-MS age; Kylander-Clark et al., 2013) was used as a primary reference for trace-element data reduction.Zircon was analyzed in two sessions: the first using a 24.1 µm spot diameter and the second using a 19.3 µm spot diameter, with 4 Hz frequency, laser energy of 3 mJ, and 25 s analysis period.Zircon analyses targeted both core and rim domains as observed in CL images.Unknowns were bracketed by primary reference zircon 91500 (1062.4± 0.4 Ma 206 Pb/ 238 U ID-TIMS age; Wiedenbeck et al., 1995) and secondary reference zircon SL-1 (563.5 ± 3.2 Ma 206 Pb/ 238 U ID-TIMS age; Gehrels et al., 2008).
Data reduction was carried out using Igor Pro v. 6.22A with Iolite v. 2.15 software (Paton et al., 2010).During all analytical sessions, repeat analyses of FC-1 gave a weighted mean 206 Pb/ 238 U age of 56.2 ± 0.1 Ma (n = 99) and a weighted mean 208 Pb/ 232 Th age of 54.0 ± 0.1 Ma (n = 99); repeat analyses of SL-1 gave a weighted mean 206 Pb/ 238 U age of 565.4 ± 2 Ma (n = 42).The Isoplot Excel plugin, version 3.75, was used to calculate weighted mean ages and plot U-Pb data (Ludwig, 2012).Uncertainties on individual analyses are quoted at the 2σ confidence level and include excess error based on the ability of the primary reference materials to accurately reproduce secondary reference materials (typically ~1% added in quadrature).Because of the high Th/U ratios in monazite, the 208 Pb/ 232 Th age is relatively insensitive to common Pb compared to 206 Pb/ 238 U ages, with typical corrections being on the order of 1 m.y.; therefore, no correction for common Pb was applied.All monazite shows some degree of reverse discordance on 206 Pb/ 238 U-208 Pb/ 232 Th concordia plots.Reverse discordance appears to be greater for monazite with higher Th/U, indicating that unsupported 206 Pb produced by the decay of 230 Th is present.Unsupported 206 Pb produces 206 Pb/ 238 U dates that are too old (Schärer, 1984).We address this issue by reporting 208 Pb/ 232 Th monazite ages as opposed to 206 Pb/ 238 U ages, on the assumption that all discordance is the result of unsupported 206 Pb.A 207 Pb-based correction (Andersen, 2002) was applied to zircon data using Isoplot v. 2.4 and the common-Pb composition derived from the single-stage model of Stacey and Kramers (1975) at the inferred crystallization age.The uncertainty on the 207 Pb-corrected date incorporates uncertainties on the measured 206 Pb/ 238 U and 207 Pb/ 206 Pb ratios as well as a 1% uncertainty on the assumed common-Pb composition.

Whole-Rock Geochemistry
Whole-rock geochemistry results are presented in Table 2 and summarized in Figure 4.The pegmatite samples share similar major-element compositions, with a restricted range in SiO 2 of 73.9-79.2wt%.Based on the alumina saturation index (Fig. 4A), the pegmatites are defined as S-type, peraluminous granites.Rb/Sr ratios are <3, and 9 of the 12 samples analyzed are similar to whole-rock Rb/Sr values from the Alpine Schist (Roser and Cooper, 1990).When plotted against Ba concentration, the majority of samples fall along a horizontal trend parallel to the Alpine Schist, while three outliers have higher Rb/Sr ratios (Fig. 4B).All samples except MR13-29 show high normative albite (Ab) and overlap the fields for H 2 O-saturated melting of dacite/graywacke (Conrad et al., 1988) and H 2 O-present melting of mica schist (Patiño Douce and Harris, 1998) when plotted on a ternary anorthite-albite-orthoclase (An-Ab-Or) diagram (Fig. 4C).MR13-29 is significantly higher in Or, while other samples show an increase in Ab over time (Fig. 4C).Chondrite-normalized whole-rock REE abundances vary by up to two orders of magnitude between samples (Fig. 4D).Half of the samples have low Eu/Eu* values of 0.1-0.8,while the rest show high Eu/Eu* values of 0.8-4.6,where the europium anomaly was calculated as Eu/([Sm + Gd] 0.5 ).Those samples with the highest Eu/Eu* values contain the highest K 2 O and Ba and lowest SiO 2 contents.

LASS U-Pb Geochronology and Trace-Element Geochemistry
Monazite, xenotime, and zircon ages are summarized in Table 1.Monazite and xenotime U-Th/Pb and REE data are summarized in Figure 5, along with X-ray maps and laser spot locations for representative crystals.Zircon geochronology and REE patterns are shown in Figure 6.Plots of monazite trace-element geochemistry are illustrated in Figure 7. U-Pb and trace element data for all accessory mineral analyses are provided in the GSA Data Repository Item1 .
MR13-3 was collected from a pegmatite sill also from Doughboy Creek (Fig. 1).Monazite in this sample displays a combination of sector and oscillatory zoning, with irregular HREE-rich cores, HREE-depleted mantles, and thin Y-rich rims (Fig. 5B).Monazite 208 Pb/ 232 Th dates define two distinct age populations showing the same trace-element signatures to those of DC2 (Fig. 5B).Older dates range from 57.7 to 55.5 Ma (n = 19), and younger dates define a weighted mean age of 54.9 ± 0.2 Ma (MSWD = 1.1; n = 21).The variation in 208 Pb/ 232 Th dates occurs between different monazite crystals, with single crystals yielding statistically similar dates.
Monazite in this sample is irregularly shaped, with grain boundaries truncating sector zoning (Fig. 5C).Xenotime is euhedral and exhibits patchy, sector, or concentric zoning with U-rich inclusions and heterogeneous trace-element distributions (Fig. 5C).Monazite 208 Pb/ 232 Th dates define a single population of 71.8 ± 0.2 Ma (MSWD = 1.5; n = 28).No correlations between monazite age and REE distribution were observed.Xenotime 206 Pb/ 238 U dates define a broader distribution than monazite, ranging from 72.8 to 63.7 Ma (Fig. 5C).The younger ages correspond to both the highest and lowest U concentrations, and they are from xenotime containing U-rich inclusions (Fig. 5C).

Sample MR13-12
Monazite in this sample is large (~75-200 µm in diameter) and euhedral to subhedral, and it exhibits sector zoning (Fig. 5D).Xenotime contains abundant U-rich inclusions and shows irregular, patchy zoning in U and Th, and homogeneous REE patterns (Fig. 5D).Monazite yields a continuous spread in 208 Pb/ 232 Th dates, ranging from 52.4 to 49.3 Ma, and the youngest analyses define a weighted mean age of 49.8 ± 0.2 Ma (MSWD = 1.1; n = 12).Minor core-to-rim age variation is resolvable in three out of eight monazite crystals.Older core monazite domains are associated with higher HREE abundances compared to the younger rims.Xenotime cores with abundant U-rich inclusions yield discordant 206 Pb/ 238 U dates, ranging from 45.7 to 55.3 Ma (n = 14; Fig. 5D).The remaining dates define a weighted mean 206 Pb/ 238 U age of 51.0 ± 0.2 Ma (MSWD = 1.2; n = 25), within the range of monazite dates.

Sample MR13-14
Monazite in this sample ranges from ~100 to 500 µm in diameter, and larger 1 mm monazite was observed in hand specimen prior to crushing.The smaller size fraction exhibits a combination of sector and oscillatory zoning, while larger fraction is more compositionally homogeneous and exhibits sector zoning only (Fig. 5E).Xenotime averages ~400 µm in diameter and exhibits oscillatory zoning defined by U and Th, which decrease in concentration toward rims (Fig. 5E).Xenotime crystals are inclusion-free, and there is little variation in REE concentrations, which display a negative Eu anomaly (Fig. 5E).Zircon exhibits patchy zoning, with no distinct cores or rims observed (Fig. 6C).CL-bright zircon domains correspond with discordant 206 Pb/ 238 U dates, indicating minor recrystallization and/or redistribution of radiogenic Pb (Fig. 6A).
Monazite from both size fractions yielded a continuous spread in 208 Pb/ 232 Th dates, ranging from 60.1 to 57.9 Ma, with the youngest analyses defining a weighted mean age of 58.5 ± 0.2 Ma (MSWD = 1; n = 27; Fig. 5E).Large monazite crystals appear to be homogeneous with respect to age, whereas two smaller oscillatory-zoned crystals yield older 59.8 ± 0.7 Ma cores and younger 57.8 ± 0.8 Ma rims.Younger ages have correspondingly higher HREE abundances.Inclusion-free xenotime crystals yield a weighted mean 206 Pb/ 238 U age of 59.0 ± 0.2 Ma (MSWD = 1.1; n = 34), consistent with the monazite from this sample.Zircon 206 Pb/ 238 U dates range from 60.0 to 51.8 Ma, with a dominant mode at ca. 60 Ma (Fig. 6A).

RESEARCH
A progressive decrease in age toward the rim is evident in the concentrically zoned crystals (Fig. 5F, lower crystal).One compositionally homogeneous monazite crystal yielded a weighted mean 208 Pb/ 232 Th age of 75.4 ± 1.2 Ma (MSWD = 0.05; n = 4).

Sample MR13-22
Monazite in this sample is anhedral to subhedral and exhibits relatively simple sector zoning (Fig. 5G).Monazite dates yield a single population with a 208 Pb/ 232 Th weighted mean age of 79.5 ± 0.4 Ma (MSWD = 1; n = 35).HREE concentrations vary by almost two orders of magnitude and form no clear trends (Fig. 5G).

Sample MR13-27
Monazite in this sample exhibits significant compositional heterogeneity, defined by irregular, patchy zoning and embayed core domains (Fig. 5I).Zircon crystals older than 100 Ma are much brighter in CL than younger zircon crystals and occur as stubby crystals exhibiting sector or oscillatory zoning with darker oscillatory-zoned mantles (Fig. 6C).In contrast, zircon crystals younger than 100 Ma are elongate and are characterized by convolute cores and thin, homogeneous rims.
Monazite yields a semicontinuous range of 208 Pb/ 232 Th dates from 85.0 to 74.8 Ma (Fig. 5I).Zircon yields 206 Pb/ 238 U dates that define at least three broad age groups: three analyses ranging 554.0-489.9Ma, an intermediate group of 11 analyses ranging 283.8-242.4Ma, and a young group of 11 analyses ranging 83.7-64.8Ma (Fig. 6A).Analyses that fall between the three age groups likely represent sampling of multiple age domains.

Sample MR13-29
MR13-29 monazite is small, ranging from ~60 to 150 µm, and exhibits irregular Y-rich cores and sector zoning in some crystals (Fig. 5J).Six anomalously young analyses range from 69.1 to 59.5 Ma and have low Th/U and HREE concentrations.The remaining 30 monazite analyses range from 70.2 to 74.3 Ma, with the youngest analyses defining a weighted mean age of 70.9 ± 0.2 Ma (MSWD = 1; n = 28).With the exception of the anomalously young monazite, there is no obvious correlation between age and REE distribution (Fig. 5J).

Sample MR13-30
Analyses from a single monazite crystal from an Alpine Schist sample collected 50 m from any pegmatite outcrop in the Mount Kinnaird area produced a weighted mean 208 Pb/ 232 Th age of 71.9 ± 0.4 Ma (MSWD = 0.4; n = 5), and a single outlier of 64.1 ± 1.9 Ma (Fig. 5L).The REE signature of this monazite is distinct from that of monazite separated from pegmatites, with a shallower HREE gradient and lower Gd/Yb value (Figs.5L and 7B).

LASS Summary
Collectively, monazite yields a broad range in 208 Pb/ 232 Th dates from 85 to 50 Ma (Fig. 8).Some samples (e.g., MR13-05 and MR13-26) contain a narrow population of monazite dates, while others display a broad, apparently continuous range of dates (e.g., .None of the samples spans the entire range defined by all monazite analyses (Fig. 8).Monazite crystals that define narrow, single-age distributions tend to display sector zoning (MR13-5, , while broad age distributions are correlated with oscillatory zoning and patchy or irregular core-rim domains (DC2, .Older monazite tends to have lower Y concentrations and steeper HREE profiles, defined by higher Gd/ Yb values (Figs. 5,7A,and 7B).Monazite shows a weakening Eu anomaly over time, with the exception of the youngest sample, MR13-12, which has similar Eu/Eu* values to the oldest samples (Fig. 7C).
Xenotime 206 Pb/ 238 U dates match monazite dates from the same sample, except where xenotime has patchy compositional zoning and U-rich inclusions.These xenotime crystals yield 206 Pb/ 238 U dates that are younger than monazite, indicating that in these samples, xenotime was probably modified by later recrystallization processes.Zircon 206 Pb/ 238 U dates are similar to monazite dates from the same sample yet trend toward younger dates in MR13-14 and MR13-27 and older dates in MR13-27.

Significance of Monazite Dates
The majority of samples in this study have accessory phase U-Th/Pb ages that define a semicontinuous range that is inconsistent with crystallization during a single event.Interpretation of the geologic significance of this spread in ages requires consideration of a variety of processes that accessory phases in anatectic melts may reflect, including: (1) mechanical mixing; (2) thermally mediated diffusional Pb loss; (3) primary igneous crystallization; or (4) recrystallization by dissolution-reprecipitation, due to either fluid-induced alteration or entrainment of inherited components into a later melt (e.g., Lederer et al., 2013;Cottle et al., 2018).A-type magmatism (Mortimer and Cooper, 2004;Cooper and Ireland, 2015) garnet

RESEARCH
Mechanical mixing due to the sampling of multiple distinct age domains in a single analysis yields geologically meaningless, mixed ages.For zircon, mechanical mixing likely contributed to the spread in ages in sample MR13-27, as inherited core domains were observed in CL images (Fig. 6).For monazite and xenotime, we consider mechanical mixing unlikely given the small volume sampled (9.7-µm-diameter by 4-µm-deep spots), the fact that spot locations were carefully selected using compositional maps, and because raw time-resolved isotopic and trace-element signals were carefully screened for down-hole variations.
The high closure temperature of these accessory minerals, estimated to be >900 °C (Schmitz and Bowring, 2003;Cherniak and Watson 2001;Cherniak et al., 2004), indicates that Pb loss by thermally mediated volume diffusion is unlikely to have affected these samples, which reached maximum temperatures of ~680 °C (Wallace, 1974;Grapes, 1995;Beyssac et al., 2016).
Recrystallization of accessory minerals due to fluid-induced alteration or incorporation of inherited components into a later melt often results in redistribution of radiogenic Pb and patchy compositional zoning or irregular compositional domains (Parrish, 1990;Geisler et al., 2007;Seydoux-Guillaume et al., 2012;Williams et al., 2011).Monazite from samples DC2, MR13-3, MR13-16, MR13-18, MR13-27, and MR13-29 shows compositionally distinct core-rim domains, which often correspond to resolvable variation in 208 Pb/ 232 Th dates (e.g., Fig. 5K).We infer that the U-Th/Pb system was partially reset in these samples during recrystallization event(s), which could include the incorporation of existing monazite into a later melt, or fluid-assisted dissolution-reprecipitation.Xenotime containing U-rich inclusions yields dispersed U-Pb dates that reflect redistribution of radiogenic Pb during recrystallization processes.Zircon yields dispersed ages that consistently extend to younger ages than monazite from the same sample.This could reflect preferential recrystallization of zircon as a result of inheritance, or changes in temperature and/or H 2 O concentration (Cottle et al., 2018).
Melts can incorporate inherited accessory minerals either via remelting of older melt products or inheritance from the Alpine Schist source or host rock.It is important to distinguish between these two processes; in the former scenario, accessory mineral dates reflect the igneous (re) crystallization history of the pegmatites, whereas in the latter scenario, dates reflect the history of the source or host rock.Incorporation of accessory minerals from a preexisting pegmatite due to introduction of fluid/ melt at temperatures above the wet solidus likely affected MR13-27, which was sampled from a pegmatite dike that crosscuts sills and yields broad monazite and zircon age ranges.Inheritance of zircon from the Alpine Schist source or host rock is indicated by the presence of older than 100 Ma detrital zircon cores (MR13-27); however, inheritance of monazite is more difficult to determine because it requires comparison with monazite-bearing Alpine Schist, which is extremely rare.Previously described monazite-bearing Alpine Schist includes a sample collected within 10 m of a pegmatite dike in the Mataketake Range dated at 71 ± 2 Ma (Mortimer and Cooper, 2004), and two 50-µm-diameter monazite crystals from monazite-apatite-allanite-epidote corona structures near Fox Glacier (~50 km north of the Mataketake Range) dated at 83 Ma (Scott et al., 2015).In this study, monazite was found in Alpine Schist sampled 50 m from any pegmatite outcrop in the Mount Kinnaird area, which produced a date of 71.9 ± 0.4 Ma (MSWD = 0.4).The distinctly shallower HREE gradient of this Alpine Schist monazite compared to monazite separated from pegmatites (Figs.5L and 7) suggests that such monazite, if incorporated by partial melts, was not preserved as an inherited component in the pegmatites.However, as a single monazite crystal cannot be considered representative of the metamorphic monazite present in the melt source or host rocks, additional data are required in order to determine the relative contribution of inheritance to the age distribution of samples DC2,  Samples collected from pegmatite sills located in the same vicinity and along strike from one another appear to show similar degrees of monazite recrystallization (e.g.,  are pristine, showing undisturbed sector zoning; Fig. 1).This suggests that the fluids responsible for melting, remelting, or fluid-assisted dissolution-reprecipitation of monazite were limited in extent, and perhaps channelized by the well-defined foliation in the Alpine Schist.
In summary, for monazite that has multiple compositionally distinct domains (samples DC2, MR13-3, MR13-16, MR13-18, MR13-27, and MR13-29), dates likely represent a combination of primary pegmatite crystallization, remelting or assimilation, and/or late-stage fluid alteration event(s).For these samples, it is difficult to determine whether monazite ages predate or postdate crystallization of the pegmatite itself.For the remaining samples, where monazite has not experienced significant recrystallization, dates either comprise single age populations defined by an MSWD of ~1, or narrow age ranges with MSWD >1.In these unmodified samples, the weighted mean age of the youngest subpopulation of analyses that define a single age population is interpreted as the pegmatite crystallization age (samples .The total age range of pegmatite crystallization defined by demonstrably igneous, unmodified monazite is 79.5 ± 0.4 Ma (MSWD = 1; n = 35; MR13-22) to 49.8 ± 0.2 Ma (MSWD = 1.1; n = 12; MR13-12), and this is comparable to the age range of monazite that may have experienced inheritance from previous melts and/or alteration, which is 85.0 Ma (MR13-27) to 54.1 Ma (MR13-3).These results indicate that the generation of melts occurred over a period of ~30 m.y., which is a significantly longer period of time than previously recognized.

Monazite Compositional Trends
Altered monazite shows a sharp increase in Y and HREE contents (defined by decreasing Gd/Yb ratios) from 85 Ma to 75 Ma, followed by a decrease in Y content and relatively stable HREE content from 75 to 50 Ma (Figs. 7A and 7B).In addition, from ca. 75 Ma onward, altered monazite consistently shows HREE depletion (higher Gd/Yb) compared to unaltered monazite of the same age (Fig. 7B).The transition in the compositional trends of altered monazite that occurs at ca. 75 Ma likely reflects the onset of geochemical alteration via fluid-assisted recrystallization, which resulted in HREEs and Y being more compatible in other phases (Zhu and O'Nions, 1999).
Unmodified monazite from samples with narrow age ranges (e.g., MR13-22) shows a sharp increase in Y and HREE contents from 85 Ma to 75 Ma, followed by a more subtle increase from 75 Ma to 50 Ma (Figs. 7A  and 7B).Eu/Eu* increases from ca. 85 to ca. 70 Ma, followed by a decrease from ca. 70 to ca.50 Ma (Fig. 7C), reflecting changes in feldspar abundance or fractionation over time.There are at least four possible interpretations for the observed Y and HREE trends in unmodified monazite through time: an increase in temperature, decreased competition for Y and HREEs with cocrystallizing minerals, progressive changes in melt composition due to breakdown of garnet and/or amphibole in the source, and/or progressive changes in melt composition due to changes in melt-producing reactions.Equilibrium partitioning of REEs and Y between coexisting monazite and xenotime is temperature dependent (Gratz and Heinrich, 1997), with incorporation of higher concentrations of Y and HREEs in monazite at higher equilibration temperatures.Although Y content in monazite can also be influenced by Th content, causing discrepancies in the Y values in the monazite-xenotime thermometer (Seydoux-Guillaume et al., 2002), Th and Y concentrations are not correlated in these monazite samples (Fig. 7D).However, the increase in HREE concentration in monazite due to an increase in temperature alone should be no more than approximately one order of magnitude (Pyle et al., 2001), whereas the observed variation in Y reported here is approximately two orders of magnitude, and so it seems unlikely that these trends are solely due to increasing temperature.
Cocrystallizing minerals such as garnet and xenotime are important controls on monazite REE distribution (Zhu and O'Nions, 1999), yet modal garnet and xenotime proportions in pegmatites do not correlate with the monazite Y and HREE trends.Furthermore, the samples with the highest Y and HREEs in monazite (MR13-5,  were the only samples to contain significant proportions of xenotime for dating. Progressive breakdown of a HREE-bearing mineral such as garnet, xenotime, or amphibole in the melt source would result in an increase in Y and HREEs both in the melt and in monazite that equilibrated with that melt.Garnet has been dated from Alpine Schist in the nearby locations of Marks Range (43.96°S, 169.11°E),Haast River (43.95°S, 169.34°E), and Karangarua River (43.59°S, 169.81°E), producing Lu-Hf garnet ages of 91.8 ± 0.3 Ma (MSWD = 1.4), 97.3 ± 0.3 Ma (MSWD = 1.2), and 78.4 ± 0.5 Ma (MSWD = 2.7), respectively (Fig. 1; Briggs et al., 2018).An amphibolite schist from the Haast River produced a Sm-Nd isochron age of 100 ± 12 Ma (Fig. 1; Mortimer and Cooper, 2004).Although garnet has not been dated from the Mataketake Range, based on these nearest garnet ages, it is possible that the monazite Y and HREE trends represent the beginning of garnet breakdown at ca. 80 Ma in this region.Garnet from 200 km north of the Mataketake Range (Fig. 1) produced a Lu-Hf age of 75.4 ± 1.3 Ma (MSWD = 0.6; Briggs et al., 2018).If the HREE trends in monazite represent garnet breakdown, this suggests that garnet-unstable conditions were restricted to the southern regions of the Alpine Schist, while garnet growth persisted in the northern Alpine Schist.Such a scenario is supported by multiple lines of evidence suggesting that during the Cretaceous, widespread horizontal extension and exhumation occurred concurrently with ongoing horizontal shortening and accretion in the Alpine Schist region of Zealandia (e.g., Vry et al., 2004;Cooper and Ireland, 2015;Briggs et al., 2018;Cooper and Palin, 2018, and references therein).
Alternatively, the trends in monazite Y and HREEs could reflect changes in the melt-producing reactions.Some dehydration melting and waterpresent melting reactions are capable of producing peritectic garnet (e.g., Weinberg and Hasalová, 2015), which retain HREEs and Y.In this scenario, increasing HREE and Y concentrations in monazite over time could reflect a progressive decrease in peritectic garnet over time.The melting reactions that produced pegmatites are discussed further below.We therefore interpret the increase in Y and HREE contents in monazite over time to reflect the breakdown of garnet, xenotime, or amphibole in the melt source, or a decrease in peritectic garnet formed during melting.

Melt Generation
Zircon that records pegmatite crystallization has radiogenic ε Hf (t) signatures ranging from 0 to +11.This range is consistent with derivation of the melts from the Alpine Schist, which has whole-rock ε Hf (t) values of −3 to +9 (Briggs et al., 2018).Rb/Sr and Ba concentrations of most samples are consistent with pegmatites being produced by H 2 O-present muscovite-breakdown melting of the Alpine Schist (Fig. 4B; Inger and Harris, 1993).The exceptions are two of the older pegmatite samples, MR13-22 and MR13-29, which have higher Rb/Sr values suggesting they were produced by H 2 O-absent dehydration melting.When plotted on a ternary An-Ab-Or diagram (Fig. 4C), the spread in samples between An and Or either indicates variable source-rock composition or variable water activity (aH 2 O); at high aH 2 O, melts become richer in Ab and An and poorer in Or (Weinberg and Hasalová, 2015).MR13-29 has a high Or content, indicating that water activity during melting was low.Thus, based on the assumption that the composition of the melt source remained relatively constant over time, geochemical signatures suggest that waterfluxed melting was dominant, with several earlier melts potentially formed either by H 2 O-absent dehydration melting or at low aH 2 O (Fig. 4C).Based on the metagraywacke composition of the melt source, possible melting reactions include: qtz + pl(ab) + kfs(or) + H 2 O = melt; ms + pl + qtz + H 2 O = melt; gt + bt + H 2 O = gt + bt + melt; and/or qtz + kfs + pl + H 2 O = pl + melt (mineral abbreviations from Kretz, 1983;Johnson et al., 2008;Weinberg and Hasalová, 2015, and references therein).

Relationship Between Metamorphism and Melting
The pegmatite emplacement ages reported here provide new insights into the relationship between melting and metamorphism in the Alpine Schist.Existing ages of prograde garnet growth in the Alpine Schist are diachronous, ranging from 100 Ma in the south to 75 Ma in the north (Fig. 1; Mortimer and Cooper, 2004;Vry et al., 2004;Briggs et al., 2018).Existing accessory mineral dates from the Alpine Schist are not yet reliably linked to the metamorphic conditions at which they formed, and it is unclear whether they record high-temperature metamorphic growth or low-temperature hydrothermal (re)crystallization.Thus, for the purpose of assessing the temporal relationship between metamorphism and melting in the Alpine Schist, garnet ages are considered to be the most reliable available record of the timing of metamorphism.Partial melting of the Alpine Schist in the Mataketake Range area initiated at 80 Ma and continued until 50 Ma, coinciding with the youngest record of garnet growth in this region at 78.4 ± 0.5 Ma and persisting for ~28 m.y. after this time (Fig. 8; Briggs et al., 2018).We infer that the semicontinuous generation of melt over such a prolonged period following metamorphism is better explained by waterfluxed melting as opposed to dehydration melting.Water-fluxed melting requires periodic infiltration of fluids into the Alpine Schist, which achieved temperatures of at least ~650 °C, either continuously or during multiple intervals over a 30 m.y.period.Such protracted periods of melting have been observed in high-temperature migmatite terranes (e.g., Montero et al., 2004), but they are less commonly observed in amphibolite-facies terranes (e.g., Rubatto et al., 2009).Unlike dehydration melting, which requires higher temperatures and can be maintained only until hydrous minerals have been consumed, water-fluxed melting can occur repeatedly in rocks that lie at, or above, the wet-melting solidus (~650 °C) whenever fluid enters the system (e.g., Rubatto et al., 2009).At present, there is no direct evidence for a long history of elevated temperatures (>650 °C), or significant temperature increases, in the Alpine Schist after 75 Ma.However, high temperatures and periodic fluid fluxing could have affected deeper, unexposed levels of the Alpine Schist during 80-50 Ma, generating melt that then migrated to presently exposed levels in the Mataketake Range.
During the 80-50 Ma period of partial melting of the Alpine Schist, Zealandia was undergoing regional extension and rifting from East Gondwana via opening of the Tasman Sea and Southern Ocean (Figs. 8 and 10;Gaina et al., 1998).In this regional tectonic context, one potential mechanism that could account for periodic recharge of fluids into the Alpine Schist is intraplate deformation focused in a narrow zone between the NW and SE regions of Zealandia (Fig. 10; Mortimer, 2014Mortimer, , 2018;;Matthews et al., 2015;Lamb et al., 2016;van der Meer et al., 2016;Cooper and Palin, 2018).Mortimer (2018) summarized a range of structural, metamorphic, and magmatic events that occurred during the Late Cretaceous to Paleogene, consistent with distributed deformation on a major crustal shear between the NW and SE Zealandia blocks.The precise nature of this Late Cretaceous-Paleogene zone of intraplate deformation remains unresolved.Cooper and Palin (2018) suggested that this NW-SE Zealandia boundary could be where Pounamu terrane sediments were accreted to the west side of the SE Zealandia microplate.Others have inferred this boundary to be a potential precursor to the currently active Alpine fault plate boundary (Sutherland et al., 2000;van der Meer et al., 2016;Mortimer, 2018).Based on concurrent seafloor spreading in the Tasman Sea, it has been suggested that this zone of intraplate deformation was accommodated on a northeast continuation of the Emerald fracture zone, which plate reconstructions show to pass near the Mataketake Range region of the South Island (Fig. 10; Schellart et al., 2006;Matthews et al., 2015).Areas of localized deformation such as shear zones are typically characterized by increased permeability, a common driver for water-present partial melting (Mogk, 1992;Mancktelow, 2006;Genier et al., 2008;Weinberg and Mark, 2008;Gordon et al., 2009;Hasalová et al., 2011).Thus, increased permeability resulting from intraplate deformation during the Late Cretaceous to Paleogene is proposed as a potential driver for water-present partial melting over a 30 m.y.period in the Alpine Schist.over a period of ~30 m.y., from 80 to 50 Ma.These results extend the known lower time limit for pegmatite emplacement by 17 m.y.Monazite trace-element compositions are consistent with gradual enrichment of Y and HREEs in melts over time, resulting from the progressive breakdown of garnet, xenotime, or amphibole in the source schist, or a decrease in peritectic garnet formed during melting.Pegmatite whole-rock geochemistry suggests that melts were primarily generated by water-fluxed melting, with minor dehydration melting potentially contributing to earlier stages of melting.Partial melting of the Alpine Schist initiated immediately prior to the latest stages of garnet growth in the Mataketake Range region and persisted for ~28 m.y. after this time.These results are consistent with temperatures of at least 650 °C being achieved either continuously or during multiple intervals over a 30 m.y.period in the Alpine Schist, with short-lived melting events driven by episodic water fluxing.The 30 m.y.period of partial melting coincided with widespread extension and rifting of Zealandia from Gondwana.In this tectonic context, we suggest that intraplate deformation focused between the NW and SE regions of Zealandia during the Late Cretaceous to Paleogene represents a potential mechanism for this prolonged period of partial melting.

Figure 1 .Figure 2 .
Figure 1.(A) Map of the Mataketake Range with sample locations from this study and previous studies indicated.Metamorphic isograds are based on the QMAP Geological Map of New Zealand (Rattenbury and Isaac, 2012).Inset shows the location of the study area within the South Island of New Zealand and existing Alpine Schist geochronology.Grt-garnet, Mnz-monazite, Zrn-zircon, Kspar-K-feldspar.(B) Schematic cross section across the South Island (transect shown in inset in A).

Figure 5 .
Figure 5. Monazite and xenotime geochronology and trace-element results by sample, with X-ray maps of Y for representative monazite and U for xenotime (warmer colors indicate higher concentration; color scale varies between crystals).Monazite concordia diagrams are colored by Th/U; xenotime concordia insets are colored by U. Chondrite-normalized rare earth element (REE) plots are colored by age.MSWD-mean square of weighted deviates.(Continued on following 2 pages.) Figure 5 (continued ).

Figure 7 .
Figure 7. Monazite composition vs. age for all analyses from all samples.(A) Mole fraction Y vs. age, assuming a total cationic site occupancy of 1, is inferred to reflect a decrease in the influence of garnet over time, as is Gd/Yb.(B) Gd/Yb vs. age, showing relative enrichment in heavy rare earth elements (HREEs) over time.(C) Europium anomaly vs. age.(D) Mole fraction Y vs. mole fraction Th assuming a total cationic site occupancy of 1; the lack of correlation indicates that the Y content in these monazite samples is not a function of Th content.

Figure 8 .
Figure 8.Time line summarizing accessory mineral geochronology from pegmatites in this study; existing pegmatite geochronology; existing garnet, monazite, and zircon geochronology from the Alpine Schist; and low-temperature chronology from the Otago Schist, compared with major tectonic and magmatic events occurring in the Zealandia region from 110 Ma.Note that the length of the bars indicates ages and their uncertainties.Timing of arc magmatism is from Tulloch and Kimbrough (2003); A-type magmatism is from Tulloch et al. (2009); metamorphic core complex (MCC) formation is from Forster and Lister (2003), Kula et al. (2007), and Tulloch et al. (2009); thrust faulting is from Mazengarb and Harris (1994); mafic dike emplacement is from Waight et al. (1998) and van der Meer et al. (2013, 2016); sedimentary basin formation is from Bishop (1992) and Strogen et al. (2017); seafloor spreading is from Gaina et al. (1998); and proto-Alpine fault intraplate deformation is from Mortimer (2018).SHRIMP-sensitive high-resolution ion microprobe; ID-TIMS-isotope dilution-thermal ionization mass spectrometry.

Figure 9 .
Figure 9. ε Hf (t) vs. 206 Pb/ 238 U age (Ma) plots for four zircon-bearing pegmatites.(A) All samples with depleted mantle reference value.(B) Plot excludes zircon older than 200 Ma.Error bar on right side of plots indicates typical 1 standard deviation uncertainties.Yellow boxes highlight the range of ε Hf (t) values of peridotite xenoliths from New Zealand (Scott et al., 2016).

CONCLUSIONS
Monazite, xenotime, and zircon petrochronology from pegmatites indicates that anatectic melting of the Alpine Schist occurred repeatedly

F
is indicated by yellow star.AU-Australia; EFZ-Emerald fracture zone; HP-Hikurangi Plateau; NI-North Island; SI-South Island; SO-Southern Ocean; PO-Pacific Ocean; TS-Tasman Sea.

TABLE 1 .
SAMPLE LOCATIONS, MATAKETAKE RANGE, NEW ZEALAND †Axial plane of folded dike.

TABLE 2 .
PEGMATITE WHOLE-ROCK GEOCHEMISTRY, MATAKETAKE RANGE, NEW ZEALAND Downloaded from https://pubs.geoscienceworld.org/gsa/lithosphere/article-pdf/4618456/169.pdf by guest Pb/ 232 Th dates from this sample define the largest within-sample range of 83.0-66.7 Ma.The range consists of three distinct age populations: The oldest group ranges from 83.0 to 79.63 Ma, an intermediate group ranges from 77.7 to 73.8 Ma, and the youngest group forms a single population with a weighted mean age of 67.7 ± 0.4 Ma (MSWD = 1; n = 9).The oldest population consists of Y-poor cores with steep HREE patterns.The intermediate age population represents Y-rich domains with relatively shallow HREE profiles and a weaker negative Eu anomaly, while the youngest age population is associated with rims with variable Y concentrations and steep HREE profiles.