Petrogenesis of the Nashwaak Granite, West-Central New Brunswick, Canada: Evidence from Trace Elements, O and Hf Isotopes of Zircon, and O Isotopes of Quartz

: The petrogenesis of the Pridoli to Early Lochkovian granites in the Miramichi Highlands of New Brunswick, Canada, is controversial. This study focuses on the Pridoli Nashwaak Granite (biotite granite and two-mica granite). In situ trace elements and O and Hf isotopes in zircon, coupled with O isotopes in quartz, are used to reveal its magmatic sources and evolution processes. In the biotite granite, inherited zircon cores have broadly homogenous δ 18 O Zrc ranging from + 6.7% (cid:24) to 7.4% (cid:24) , whereas magmatic zircon rims have δ 18 O Zrc of + 6.3% (cid:24) to 7.2% (cid:24) and ε Hf(t) of − 0.39 to − 5.10. The Hf and Yb / Gd increase with decreasing Th / U. Quartz is isotopically equilibrated with magmatic zircon rims. The biotite granite is interpreted to be solely derived by partial melting of old basement rocks of Ganderia and fractionally crystallized at the f O 2 of 10 − 21 to 10 − 10 bars. The two-mica granite has heterogeneous inherited zircon cores ( δ 18 O Zrc of + 5.2% (cid:24) to 9.9% (cid:24) ) and rims ( δ 18 O Zrc of + 6.2% (cid:24) to 8.7% (cid:24) ), and ε Hf (t) of − 11.7 to − 1.01. The two-mica granite was derived from the same basement, but with supracrustal contamination. This open-system process is also recorded by Yb / Gd and Th / U ratios in zircon and isotopic disequilibrium between magmatic zircon rims and quartz ( + 10.3 ± 0.2% (cid:24) ). Author Contributions: Conceptualization, D.R.L. and W.Z.; methodology, D.R.L.; software, W.Z.; validation, D.R.L., W.Z., and K.G.T.; formal analysis, W.Z.; investigation, D.R.L.; resources, K.G.T.; data curation, W.Z.; writing—original draft preparation, W.Z.; writing—review and editing, W.Z.; visualization, W.Z.; supervision, D.R.L.; administration, funding acquisition,


Introduction
The Canadian Appalachians formed by accretion of several Gondwanan microcontinents to Laurentia as a result of Early to Late Paleozoic closure of the Iapetus and Rheic oceans [1]. Three phases of orogenesis recorded by various rock suites in New Brunswick are known as Taconic, Salinic, and Acadian orogenesis. The Taconic orogeny was manifested by accretion of three oceanic and continental terranes in the peri-Laurentia realm (500-450 Ma). Closure of the Tetagouche-Exploits backarc basin along the Bamford Brook Fault in Ganderia resulted from the Salinic orogeny (450-423 Ma). This was then followed by the Acadian orogeny due to the northwest flat-slab subduction of Avalonia beneath Ganderia along the Caledonia Fault (421-400 Ma) [2]. New Brunswick is mostly underlain by Neoproterozoic Ganderian basement (sedimentary rocks and arc volcanic rocks) that is overlain by Cambro-Ordovician, quartz-rich, passive margin sedimentary rocks. Ganderia is bordered to the north by arc volcanic rocks of the peri-Laurentian Notre Dame zone, and by the Avalonian microcontinent along the Caledonian Fault to the south [3]. Although both Ganderia and Avalonia are

Geological Background
The Nashwaak Granite intruded the Cambro-Ordovician Trousers Lake Metamorphic Suite and the Ordovician McKiel Lake Granite to the north, and the Becaguimec Lake Gabbro to the west. The contact between the Nashwaak Granite and Early Devonian volcanic rocks to the south is not exposed. On its east side, the Nashwaak Granite intruded quartzose wackes, siltstones, and shales of the Cambrian-Early Ordovician Miramichi Group, and younger volcanic and sedimentary rocks of the Ordovician Tetagouche Group (Figure 1). Andalusite and cordierite are present in sedimentary rocks up to 2 km from the contact.

Geological Background
The Nashwaak Granite intruded the Cambro-Ordovician Trousers Lake Metamorphic Suite and the Ordovician McKiel Lake Granite to the north, and the Becaguimec Lake Gabbro to the west. The contact between the Nashwaak Granite and Early Devonian volcanic rocks to the south is not exposed. On its east side, the Nashwaak Granite intruded quartzose wackes, siltstones, and shales of the Cambrian-Early Ordovician Miramichi Group, and younger volcanic and sedimentary rocks of the Ordovician Tetagouche Group (Figure 1). Andalusite and cordierite are present in sedimentary rocks up to 2 km from the contact. The Nashwaak Granite has two subfacies: (1) pink, coarse-to medium-grained, equigranular to porphyritic biotite granite with a mineral assemblage of plagioclase, orthoclase, quartz, and minor biotite, grading northward into (2) fine-to medium-grained muscovite-biotite granite containing quartz, microcline, albite, muscovite, zircon, apatite, monazite, and ilmenite. The Nashwaak Granite was formed at ca. 420 Ma and is highly siliceous (69.3-81.5 wt.%), calc-alkaline, and peraluminous, with (La/Yb) N ranging from 1.9 to 13.6, and depletion of Ba, Sr, Nb, P, and Ti. The ε Nd (t) is from −2.1 to −4.2 [8].

Analytical Methods
In this study, we used the same [8] zircon samples from the Nashwaak Granite as we previously dated for their U-Pb ages. Some of the zircons show inherited cores with 206 Pb/ 238 U ages ranging from 1000 to 1945 Ma and 207 Pb/ 206 Pb ages of 1040 to 2260 Ma [8]. With guidance from cathodoluminescence (CL) images, zircon grains with overgrowth magmatic rims (identified by similar 206 Pb/ 238 U age as that sample's concordia age) were selected for in situ trace element analysis at the rims. Oxygen isotope analysis was conducted on both rims and inherited cores. Only the zircon rims used for the concordia age calculation were chosen for hafnium isotope analysis. Detailed analytical methods are listed below.

Oxygen Isotope Analysis of Zircon and Quartz
Secondary ion mass spectrometry (SIMS) oxygen isotope analysis of zircon and quartz was carried out using a Cameca IMS 1280 ion microprobe (manufactured by CAMECA, Société par Actions Simplifiée, Gennevilliers, France) at the Canadian Centre for Isotopic Microanalysis (CCIM), University of Alberta. A 133 Cs + primary beam was operated with an impact energy of 20 keV and a 2-4 nA beam current. The~12 µm diameter probe was rastered slightly during acquisition to form rectangular sputtered areas of~15 µm × 18 µm. Negative secondary ions were extracted from the sputtered area into the secondary (transfer) column by application of a 10 kV potential gradient. Transfer conditions included a 122 µm entrance slit, a 400 µm contrast aperture, and a 5 mm field aperture. The energy window utilized was 150 eV. The mass-separated oxygen isotopes were detected simultaneously in Faraday cups L'2 ( 16 O − ) and H'2 ( 18 O − ) in the multidetector array. Mass resolution (∆m/M at 10%) was typically 1950 and 2275, respectively (see [18]). For quartz, reference material S0033 was used (GeeWiz glass [19]); the measured δ 18 O value of +12.34 ± 0.05% (n = 12) agrees well with the reported value of +12.5% . Median uncertainties for the quartz reference materials and samples at 95% confidence (2σ) were ±0.17% (Table S1). For zircon, an internal reference material S0081 (UAMT1) was used; the measured δ 18 O VSMOW value of +4.81 ± 0.04% agrees well with the accepted value of +4.87% (Stern R., unpublished data). At 95% confidence, the median uncertainty for the zircon reference materials and samples is ±0. 19 and ±0.18% , respectively (Table S2).

Trace Element and Hf Isotopic Analysis of Zircon
The in situ zircon trace element analyses were conducted using a Resonetics M-50-LR 193 nm (manufactured by Resonetics at Kettering, OH, USA) Excimer laser ablation system coupled to an Agilent 7700× quadrupole inductively coupled plasma-mass spectrometer (ICP-MS) (manufactured by Agilent technologies at Santa Clara, CA, USA) at the University of New Brunswick. A spot size of 33 µm, a beam energy of 4 J/cm 2 , and an 8 Hz laser repetition rate were employed. Calibration was achieved using standard procedures [20] that included use of standard reference materials (SRM) 610 glass from National Institute of Standards and Technology (NIST) for external standardization and the stoichiometric SiO 2 content of zircon for internal standardization (Table S3).
To evaluate whether the laser ablation spot was placed on a mineral inclusion (i.e., monazite, apatite, and titanite) that was not detectable at the scale of optical and CL imaging, the spots with extremely high content of Ca, Sr, Th, and P were carefully checked. Results with strong correlation between P and (Sm/La) N , Th and (Sm/La) N , or Ca/Sr and light rare earth elements (LREE), indicating the presence of monazite or apatite, were discarded (see [21]).
In situ Hf isotope analyses of zircons were conducted using a Resolution S-155 laser-ablation system with a beam size of 50 µm and a pulse frequency of 8 Hz, coupled with a Nu Plasma II multicollector ICP-MS at the State Key Laboratory of Geological Processes and Mineral Resources, China University of Geosciences (GPMR), Wuhan. The reference standard Penglai zircon with a 176 Hf/ 177 Hf ratio of 0.282906 ± 0.000001 (2σ) was analyzed along with the samples [22]. Detailed information about the analytical method can be found in [23] (Table S4). between P and (Sm/La)N, Th and (Sm/La)N, or Ca/Sr and light rare earth elements (LREE), indicating the presence of monazite or apatite, were discarded (see [21]). In situ Hf isotope analyses of zircons were conducted using a Resolution S-155 laser-ablation system with a beam size of 50 µm and a pulse frequency of 8 Hz, coupled with a Nu Plasma II multicollector ICP-MS at the State Key Laboratory of Geological Processes and Mineral Resources, China University of Geosciences (GPMR), Wuhan. The reference standard Penglai zircon with a 176 Hf/ 177 Hf ratio of 0.282906 ± 0.000001 (2σ) was analyzed along with the samples [22]. Detailed information about the analytical method can be found in [23] (Table S4).    between P and (Sm/La)N, Th and (Sm/La)N, or Ca/Sr and light rare earth elements (LREE), indicating the presence of monazite or apatite, were discarded (see [21]). In situ Hf isotope analyses of zircons were conducted using a Resolution S-155 laser-ablation system with a beam size of 50 µm and a pulse frequency of 8 Hz, coupled with a Nu Plasma II multicollector ICP-MS at the State Key Laboratory of Geological Processes and Mineral Resources, China University of Geosciences (GPMR), Wuhan. The reference standard Penglai zircon with a 176 Hf/ 177 Hf ratio of 0.282906 ± 0.000001 (2σ) was analyzed along with the samples [22]. Detailed information about the analytical method can be found in [23] (Table S4).     Table S3) are based on the method of [24]. They also have similar Zr (39.9-43.5 wt.%) and Hf (8750-15,090 ppm) contents. The total REE content in them ranges from 539 to 2130 ppm ( Figure 3b).

CL Images and Oxygen Isotopes in Zircon and Quartz
The quartz grains in the biotite granite and two-mica granite are green and bluish red in ChromaSEM-CL ( Figure 4). They are dominantly homogeneous and composed of only one generation of primary igneous quartz. A few quartz grains show oscillatory zoning at µm scale, revealed by small-amplitude variations in CL (Figure 4d). Secondary textures in the magmatic quartz from the Nashwaak Granite include: (1) dark CL streaks and patches associated with fractures; (2) healed fractures, which do not show up in BSE images, and are likely filled by nonluminescent SiO 2 (see [26]); and (3) opened fractures.   Table S3) are based on the method of [24]. They also have similar Zr (39.9-43.5 wt.%) and Hf (8750-15,090 ppm) contents. The total REE content in them ranges from 539 to 2130 ppm ( Figure 3b).

CL Images and Oxygen Isotopes in Zircon and Quartz
The quartz grains in the biotite granite and two-mica granite are green and bluish red in ChromaSEM-CL ( Figure 4). They are dominantly homogeneous and composed of only one generation of primary igneous quartz. A few quartz grains show oscillatory zoning at µm scale, revealed by small-amplitude variations in CL (Figure 4d). Secondary textures in the magmatic quartz from the Nashwaak Granite include: (1) dark CL streaks and patches associated with fractures; (2) healed fractures, which do not show up in BSE images, and are likely filled by nonluminescent SiO2 (see [26]); and (3) opened fractures. Overall, δ 18 O of quartz shows limited or no intragranular variations in each granite sample. In the two-mica granite, the mean δ 18 OQz is +10.3 ± 0.2‰ (2σ, n = 11), which is slightly higher than that Overall, δ 18 O of quartz shows limited or no intragranular variations in each granite sample. In the two-mica granite, the mean δ 18 O Qz is +10.3 ± 0.2% (2σ, n = 11), which is slightly higher than that of the biotite granite with a mean of +9.7 ± 0.2% (2σ, n = 19). Detailed examination of magmatic quartz yielded consistent δ 18 O Qz values, regardless of distance from the rim of a grain (Table S1).
Two zircon populations were identified using SEM-CL: (1) prevalent high-CL response (bright) zircon domains (cores or whole crystals); and (2) thin, low-CL response rims ( Figure 5). Attempts were made to analyze both of these domains; however, due to the small size of the zircon crystals and the internal fracturing observed in many grains, it was not always possible to fit multiple analyses on a single crystal. Intragrain δ 18 O Zrc values from the two-mica granite are highly variable; ranging from +5.2% to 9.9% (mean +7.4% ) for the bright cores, and +6.2% to 8.7% (mean +7.1% ) for the overgrowth oscillatory rims (Figure 6a,b). The δ 18 O Zrc values either increase or decrease from core to rim, with a range of 0.3% to 2.5% . Considering that analytical uncertainties are generally <±0.2% (2σ), these intergrain and intragrain variations likely represent real oxygen isotopic heterogeneity. In contrast, zircon from the biotite granite shows generally lower δ 18 O Zrc values and limited variation between cores (+6.7-7.4% ; mean = +7.0% ) and rims (+6.3% and 7.2% ; mean = +6.8% ) (Figure 6c,d; Table S2). The core-rim variation of each grain is insignificant within a range of less than 0.4% .
Minerals 2020, 10, x FOR PEER REVIEW 6 of 13 of the biotite granite with a mean of +9.7 ± 0.2‰ (2σ, n = 19). Detailed examination of magmatic quartz yielded consistent δ 18 OQz values, regardless of distance from the rim of a grain (Table S1). Two zircon populations were identified using SEM-CL: (1) prevalent high-CL response (bright) zircon domains (cores or whole crystals); and (2) thin, low-CL response rims ( Figure 5). Attempts were made to analyze both of these domains; however, due to the small size of the zircon crystals and the internal fracturing observed in many grains, it was not always possible to fit multiple analyses on a single crystal. Intragrain δ 18 OZrc values from the two-mica granite are highly variable; ranging from +5.2 to 9.9‰ (mean +7.4‰) for the bright cores, and +6.2 to 8.7‰ (mean +7.1‰) for the overgrowth oscillatory rims (Figure 6a,b). The δ 18 OZrc values either increase or decrease from core to rim, with a range of 0.3 to 2.5‰. Considering that analytical uncertainties are generally <±0.2‰ (2σ), these intergrain and intragrain variations likely represent real oxygen isotopic heterogeneity. In contrast, zircon from the biotite granite shows generally lower δ 18 OZrc values and limited variation between cores (+6.7-7.4‰; mean = +7.0‰) and rims (+6.3 and 7.2‰; mean = +6.8‰) (Figure 6c,d; Table S2). The core-rim variation of each grain is insignificant within a range of less than 0.4‰.    (Table S1). Two zircon populations were identified using SEM-CL: (1) prevalent high-CL response (bright) zircon domains (cores or whole crystals); and (2) thin, low-CL response rims ( Figure 5). Attempts were made to analyze both of these domains; however, due to the small size of the zircon crystals and the internal fracturing observed in many grains, it was not always possible to fit multiple analyses on a single crystal. Intragrain δ 18 OZrc values from the two-mica granite are highly variable; ranging from +5.2 to 9.9‰ (mean +7.4‰) for the bright cores, and +6.2 to 8.7‰ (mean +7.1‰) for the overgrowth oscillatory rims (Figure 6a,b). The δ 18 OZrc values either increase or decrease from core to rim, with a range of 0.3 to 2.5‰. Considering that analytical uncertainties are generally <±0.2‰ (2σ), these intergrain and intragrain variations likely represent real oxygen isotopic heterogeneity. In contrast, zircon from the biotite granite shows generally lower δ 18 OZrc values and limited variation between cores (+6.7-7.4‰; mean = +7.0‰) and rims (+6.3 and 7.2‰; mean = +6.8‰) (Figure 6c,d; Table S2). The core-rim variation of each grain is insignificant within a range of less than 0.4‰.

Hafnium Isotopes in Zircon
Zircons in the two-mica granite show a large range of 176 Hf/ 177 Hf, from 0.28219 to 0.28250, corresponding to ε Hf(t) values ranging from −11.7 to −1.0, with two-stage Hf model ages from 1464 to 2135 Ma. In contrast, zircons from the biotite granite show less variation in 176 Hf/ 177 Hf ratios (0.28238 to 0.28251) and ε Hf(t) values (from −0.4 to −5.1), with more limited two-stage Hf model ages from 1427 to 1724 Ma (Figure 7; Table S4).

Hafnium Isotopes in Zircon
Zircons in the two-mica granite show a large range of 176 Hf/ 177 Hf, from 0.28219 to 0.28250, corresponding to εHf(t) values ranging from −11.7 to −1.0, with two-stage Hf model ages from 1464 to 2135 Ma. In contrast, zircons from the biotite granite show less variation in 176 Hf/ 177 Hf ratios (0.28238 to 0.28251) and εHf(t) values (from −0.4 to −5.1), with more limited two-stage Hf model ages from 1427 to 1724 Ma (Figure 7; Table S4).

Crystal Fractionation and Contamination
Trace elements incorporated into zircon reflect the composition of host magma in which it is growing. Crystal fractionation produces residual melts with relatively higher U and Th content; consequently zircon in such melts evolves toward high Th and U content and low Th/U [27]. Hafnium increases and Zr/Hf decreases in zircon with magma cooling [28][29][30]. Both the biotite granite and two-mica granite in the Nashwaak Granite have similar U and Th content (Table S3), but zircon rims in the two-mica granite have a wider range of Th/U and Hf content. Relative enrichment in HREE over middle rare earth elements (MREE) in zircons is interpreted as a result of mineral (garnet, hornblende, titanite, and apatite) fractionation in a closed system (Table S3, [17]). In the Nashwaak Granite, only apatite typically occurs as inclusions within biotite. The plot of Th/U versus Yb/Gd may differentiate crystal fractionation and mixing/contamination ( [17] Figure 2d). For example, zircon rims of the biotite granite follow the crystal fractionation curve, whereas zircon rims of the two-mica granite are distributed along a mixing curve where Th/U is higher than 0.2 (Table S3). When Th/U is lower than 0.2, zircon rims of two-mica granite plot on the crystal fractionation trend, but some of these grains have relatively high Ti contents (corresponding to high temperatures; [31]), which contradicts a gradual cooling process during fractional crystallization. Thus, zircons with low Th/U and high Ti may also be formed by crustal contamination.

Titanium-In-Zircon Thermometry
The titanium-in-zircon thermometer [28,31,32] provides zircon crystallization temperatures in host melts [17]. Two key parameters of this thermometer are the activities of SiO2 (aSiO2) and TiO2 (aTiO2) in the magma. Model temperatures are calculated for magmas with titanite, in which case a value of 0.7 is adopted for aTiO2, as suggested by [17] and [33]. Variation of 0.1 in aTiO2 leads to changes of 25 to 30 °C in the calculated temperatures. For the Nashwaak Granite, the aSiO2 is 1.0 due to the presence of quartz, and aTiO2 is 0.6, based on the absence of titanite (0.6 is the minimum for felsic magmas). According to this model, the Ti contents of magmatic rims from the biotite granite

Crystal Fractionation and Contamination
Trace elements incorporated into zircon reflect the composition of host magma in which it is growing. Crystal fractionation produces residual melts with relatively higher U and Th content; consequently zircon in such melts evolves toward high Th and U content and low Th/U [27]. Hafnium increases and Zr/Hf decreases in zircon with magma cooling [28][29][30]. Both the biotite granite and two-mica granite in the Nashwaak Granite have similar U and Th content (Table S3), but zircon rims in the two-mica granite have a wider range of Th/U and Hf content. Relative enrichment in HREE over middle rare earth elements (MREE) in zircons is interpreted as a result of mineral (garnet, hornblende, titanite, and apatite) fractionation in a closed system (Table S3, [17]). In the Nashwaak Granite, only apatite typically occurs as inclusions within biotite. The plot of Th/U versus Yb/Gd may differentiate crystal fractionation and mixing/contamination ( [17], Figure 2d). For example, zircon rims of the biotite granite follow the crystal fractionation curve, whereas zircon rims of the two-mica granite are distributed along a mixing curve where Th/U is higher than 0.2 (Table S3). When Th/U is lower than 0.2, zircon rims of two-mica granite plot on the crystal fractionation trend, but some of these grains have relatively high Ti contents (corresponding to high temperatures; [31]), which contradicts a gradual cooling process during fractional crystallization. Thus, zircons with low Th/U and high Ti may also be formed by crustal contamination.

Titanium-In-Zircon Thermometry
The titanium-in-zircon thermometer [28,31,32] provides zircon crystallization temperatures in host melts [17]. Two key parameters of this thermometer are the activities of SiO 2 (aSiO 2 ) and TiO 2 (aTiO 2 ) in the magma. Model temperatures are calculated for magmas with titanite, in which case a value of 0.7 is adopted for aTiO 2 , as suggested by [17] and [33]. Variation of 0.1 in aTiO 2 leads to changes of 25 to 30 • C in the calculated temperatures. For the Nashwaak Granite, the aSiO 2 is 1.0 due to the presence of quartz, and aTiO 2 is 0.6, based on the absence of titanite (0.6 is the minimum for felsic magmas). According to this model, the Ti contents of magmatic rims from the biotite granite broadly crystallized at 710 to 850 • C, whereas those of the two-mica granite mainly crystallized between 750 and 800 • C (Figure 8; Table S3).
Minerals 2020, 10, x FOR PEER REVIEW 8 of 13 broadly crystallized at 710 to 850 °C, whereas those of the two-mica granite mainly crystallized between 750 and 800 °C (Figure 8; Table S3).

Figure 8.
Cumulative probability histograms of the Ti-in-zircon thermometer [31] for the two-mica granite (a) and biotite granite (b).

Oxygen Fugacity
Variation of Ce and Eu anomalies in zircon (Figure 2) reflects the oxidation state of magmas [17]. Cerium and Eu have two valence states. Compared with Ce 3+ , Ce 4+ is incorporated preferentially into the Zr 4+ site of zircon; whereas Eu 2+ is easily accommodated into the Ca 2+ site of plagioclase. An increase in the oxidation state of magma enhances the positive Ce anomaly, but weakens the negative Eu anomaly in zircon [34].
The Eu/Eu* of zircon rims in the biotite granite generally decreases along with increasing Hf content and decreasing Th/U (Figure 2). Although the influence of oxygen fugacity cannot be totally precluded, the enhancement of a negative Eu anomaly might be dominantly controlled by plagioclase fractional crystallization [30]. Zircon rims in the two-mica granite generally have a Eu/Eu* broadly lower than 0.4, but some scattered values are in the range of 0.6 to 1.8 (Figure 2b). In this case, crustal contamination as discussed above may play a vital role in evolution of the two-mica granite.
An alternative method to evaluate the oxidation state of magma uses the zircon Ce anomaly. In order to avoid error in the estimation of Ce/Ce* arising from low La and Pr content in zircon, the lattice-strain model [24] was used to calculate Ce/Ce*, corresponding Ce 4+ /Ce 3+ [35], and oxygen fugacity [36] of the magmas. The zircon rims from the biotite granite have Ce 4+ /Ce 3+ in the range of 1.5 to 44.6 (mean 17.5), with calculated fO2 of 10 −21 to 10 −10 (mean 10 −15 bars) (Figure 9), whereas the zircon rims from the two-mica granite are relatively reduced with Ce 4+ /Ce 3+ of 1.7 to 24.4 (mean 10.1) and calculated fO2 of 10 −23 to 10 −13 (mean 10 −17 ) bars ( Figure 9; Table S3). An evolution of fO2 with progressive crystallization in two-mica granite is not observed, whereas the fO2 of biotite granite increases with decreasing Th/U. This oxidation state could be attributed to H2 degassing [37] or the result of reduction of sulfate to sulfur dioxide during separation of a sulfur-rich magmatichydrothermal fluid [38].

Oxygen Fugacity
Variation of Ce and Eu anomalies in zircon (Figure 2) reflects the oxidation state of magmas [17]. Cerium and Eu have two valence states. Compared with Ce 3+ , Ce 4+ is incorporated preferentially into the Zr 4+ site of zircon; whereas Eu 2+ is easily accommodated into the Ca 2+ site of plagioclase. An increase in the oxidation state of magma enhances the positive Ce anomaly, but weakens the negative Eu anomaly in zircon [34].
The Eu/Eu* of zircon rims in the biotite granite generally decreases along with increasing Hf content and decreasing Th/U (Figure 2). Although the influence of oxygen fugacity cannot be totally precluded, the enhancement of a negative Eu anomaly might be dominantly controlled by plagioclase fractional crystallization [30]. Zircon rims in the two-mica granite generally have a Eu/Eu* broadly lower than 0.4, but some scattered values are in the range of 0.6 to 1.8 (Figure 2b). In this case, crustal contamination as discussed above may play a vital role in evolution of the two-mica granite.
An alternative method to evaluate the oxidation state of magma uses the zircon Ce anomaly. In order to avoid error in the estimation of Ce/Ce* arising from low La and Pr content in zircon, the lattice-strain model [24] was used to calculate Ce/Ce*, corresponding Ce 4+ /Ce 3+ [35], and oxygen fugacity [36] of the magmas. The zircon rims from the biotite granite have Ce 4+ /Ce 3+ in the range of 1.5 to 44.6 (mean 17.5), with calculated f O 2 of 10 −21 to 10 −10 (mean 10 −15 bars) (Figure 9), whereas the zircon rims from the two-mica granite are relatively reduced with Ce 4+ /Ce 3+ of 1.7 to 24.4 (mean 10.1) and calculated f O 2 of 10 −23 to 10 −13 (mean 10 −17 ) bars ( Figure 9; Table S3). An evolution of f O 2 with progressive crystallization in two-mica granite is not observed, whereas the f O 2 of biotite granite increases with decreasing Th/U. This oxidation state could be attributed to H 2 degassing [37] or the result of reduction of sulfate to sulfur dioxide during separation of a sulfur-rich magmatic-hydrothermal fluid [38].
If quartz and zircon are in isotopic equilibrium at a particular temperature, then their δ 18 O values should lie along a straight line in a plot of δ 18 OZrc vs. δ 18 OQtz (see Figure 5b in [11]). In the biotite granite, the δ 18 OZrc value of the magmatic rim is broadly homogenous (from +6.3 to +7.2‰). Consequently, the zircon and quartz in the biotite granite might have formed in a closed system and equilibrated with their hosting magma during crystallization. The average δ 18 O values of zircon rims and quartz are +6.8 and +9.7‰, respectively. The value of ΔQtz-Zrc is 2.9‰ and the calculated temperature is 688 °C (Tables S1 and S2; [40]) or 675 °C [39].
In the two-mica granite, the zircon has heterogeneous cores with the δ 18 OZrc ranging from +5.2 to +9.9‰, indicating various magmatic sources. The zircon rims have δ 18 OZrc in the range of 6.2 to 8.7‰ with a mean value of 7.1‰. The average O-isotope fractionation between zircon rims and quartz yields an equilibration temperature below 640 °C [40] or 627 °C [39], much lower than the Ti-in-zircon temperatures. Furthermore, the 2.5‰ difference of δ 18 OZrc in the zircon rims indicates that this magma crystallized in an open system.
Oxygen isotope disequilibrium between zircon and quartz could be caused by several processes, as suggested by [16] for granitoids elsewhere. Feldspar and quartz are less refractory minerals in granites compared to zircon, therefore their oxygen isotope compositions could be readily reset by hydrothermal alteration and recrystallization [15,41,42]. However, large-scale or intense hydrothermal alteration is not evident in the Nashwaak Granite, as indicated by the quartz texture in the CL images. Although quartz grains in biotite granite rarely have dilatant fractures along which hydrothermal alteration could occur (Figure 4c), transects across whole magmatic quartz grains do not show any δ 18 OQz variation, suggesting that hydrothermal alteration was not a main process causing oxygen isotope disequilibrium between quartz and zircon.
Magma mixing (e.g., [43][44][45]), and crustal assimilation (e.g., [14][15][16]46,47]) are other processes that can change the δ 18 O composition of melts. As shown in [48], early-crystallizing zircon has lower δ 18 O than late-crystallized garnet, which formed after contamination of the granitic magma by assimilation of sedimentary rocks. Supracrustal contamination in the two-mica granite is indicated by the presence of high δ 18 O (>+8‰) zircon cores. Such a contamination process could affect the composition and oxygen fugacity of melts. That the δ 18 O of magmatic quartz in the two-mica granite

Magma Sources and Evolution as Recorded by O-Hf Isotopes
Oxygen isotope fractionation between zircon and quartz is a function of temperature and is independent of oxygen fugacity or pressure (Equation (1)): where A Qtz-Zrc = 2.33 ± 2.04 [39] or 2.64 [40] and T = temperature in Kelvin.
If quartz and zircon are in isotopic equilibrium at a particular temperature, then their δ 18 O values should lie along a straight line in a plot of δ 18 O Zrc vs. δ 18 O Qtz (see Figure 5b in [11]). In the biotite granite, the δ 18 O Zrc value of the magmatic rim is broadly homogenous (from +6.3% to +7.2% ). Consequently, the zircon and quartz in the biotite granite might have formed in a closed system and equilibrated with their hosting magma during crystallization. The average δ 18 O values of zircon rims and quartz are +6.8% and +9.7% , respectively. The value of ∆ Qtz-Zrc is 2.9% and the calculated temperature is 688 • C (Tables S1 and S2; [40]) or 675 • C [39].
In the two-mica granite, the zircon has heterogeneous cores with the δ 18 O Zrc ranging from +5.2% to +9.9% , indicating various magmatic sources. The zircon rims have δ 18 O Zrc in the range of 6.2% to 8.7% with a mean value of 7.1% . The average O-isotope fractionation between zircon rims and quartz yields an equilibration temperature below 640 • C [40] or 627 • C [39], much lower than the Ti-in-zircon temperatures. Furthermore, the 2.5% difference of δ 18 O Zrc in the zircon rims indicates that this magma crystallized in an open system.
Oxygen isotope disequilibrium between zircon and quartz could be caused by several processes, as suggested by [16] for granitoids elsewhere. Feldspar and quartz are less refractory minerals in granites compared to zircon, therefore their oxygen isotope compositions could be readily reset by hydrothermal alteration and recrystallization [15,41,42]. However, large-scale or intense hydrothermal alteration is not evident in the Nashwaak Granite, as indicated by the quartz texture in the CL images. Although quartz grains in biotite granite rarely have dilatant fractures along which hydrothermal alteration could occur (Figure 4c), transects across whole magmatic quartz grains do not show any δ 18 O Qz variation, suggesting that hydrothermal alteration was not a main process causing oxygen isotope disequilibrium between quartz and zircon.
Magma mixing (e.g., [43][44][45]), and crustal assimilation (e.g., [14][15][16]46,47]) are other processes that can change the δ 18 O composition of melts. As shown in [48], early-crystallizing zircon has lower δ 18 O than late-crystallized garnet, which formed after contamination of the granitic magma by assimilation of sedimentary rocks. Supracrustal contamination in the two-mica granite is indicated by the presence of high δ 18 O (>+8% ) zircon cores. Such a contamination process could affect the composition and oxygen fugacity of melts. That the δ 18 O of magmatic quartz in the two-mica granite did not record the input of supracrustal materials might be due to very late stage crystallization of quartz, after the magma was completely homogenized (see [43]).
Various magma sources can be identified by the oxygen isotope compositions of zircon. Magmatic zircons equilibrated with mantle-derived magmas have average δ 18 O Zrc value of +5.3 ± 0.6% (2σ; [49]). Notable deviations of δ 18 O Zrc from the mantle value are the result of intracrustal recycling. High-δ 18 O magmas (+8% to above +10% ) reflect assimilation of supracrustal rocks that previously interacted with low temperature fluids [50], while low δ 18 O magmas (lower than ca. +4% ) reflect assimilation of supracrustal rocks that previously interacted with meteoric water or seawater at high temperatures [46]. For the Nashwaak Granite, zircon in the biotite granite is in high temperature equilibrium with the coexisting quartz and crystallized in a closed system, thus hydrothermal alteration or assimilation of crustal rock can be ruled out during magma evolution. The homogeneous δ 18 O Zr within single grain or between different zircon rims, relatively homogeneous ε Hf (t) (from −0.39 to −5.10) and "I-type" whole-rock geochemistry characteristics [8] indicate that the biotite granite might be dominantly derived by partially melting of meta-igneous rocks in the lower crust of Ganderia.
The two-mica granite has heterogeneous zircon cores (+5.2% to +9.9% ), indicating multiple magma sources. Most δ 18 O Zrc of the zircon cores in the two-mica granite are the same as those of the biotite granite. Three cores with δ 18 O Zrc values of +5.23% to +5.45% represent a crustal mafic rock source instead of a juvenile mantle source because their 206 Pb/ 238 U ages are older than 1.0 Ga. Zircon cores with δ 18 O Zrc of +9.51% to +9.94% are the result of assimilation of supracrustal rocks. This is supported by the feldspar Pb isotopic compositions of two-mica granite, which plot along or near the upper crust reference curve ( 206 Pb/ 204 Pb vs. 207 Pb/ 204 Pb) and on or near the orogene reference curve ( 206 Pb/ 204 Pb vs. 208 Pb/ 204 Pb) (see [51]). The biotite granite has ε Hf (t) of −0.39 to −5.10 and older T DM2 ages of 1427 to 1724 Ma, while the two-mica granite has more negative ε Hf (t) (as low as −11.7) and older T DM2 ages (<2135 Ma). This isotopic difference, caused by crustal assimilation, indicates that an "inverted" crustal structure (low ε Nd (t) sedimentary rocks overlying more positive ε Nd (t) crust), similar to that invoked by [4] for the Ganderia in Newfoundland.

Conclusions
This study indicates the in situ analysis of trace elements of zircon as well as oxygen isotope of both zircon and quartz is an effective way to distinguish crustal contamination from heterogeneity of magmatic source. Some Pridoli to Early Lochkovian granites in Ganderia, proposed to be formed by partial melting of heterogeneous source material (e.g., [52]), might not always be the case, such as the Nashwaak Granite investigated by this work. In situ Hf and O isotope measurements in zircon, the first such data reported for granites in New Brunswick's Central Plutonic Belt, record the detailed magmatic sources and petrogenetic processes of the Nashwaak Granite in the Miramichi Highlands. This granite is dominantly derived from partial melting of Ganderian lower crust and contamination from various supracrustal rocks. The unexposed basement has δ 18 O Zrc values between +6.7% and +7.4% , as reflected by inherited zircon cores in the biotite granite. The supracrustal rocks, including mafic rocks with δ 18 O Zrc of +5.2% to +5.5% and sedimentary materials with δ 18 O Zrc of +9.5% to +9.9% , were incorporated into a magma derived from the lower basement. Assimilation of those materials led to a decrease in the oxygen fugacity of the hybrid magmas, as indicated by trace element compositions of zircon.