A Record of Paleoproterozoic Subduction Preserved in the Northern Slave Cratonic Mantle : Sr ^ Pb ^ O Isotope andTrace-element Investigations of Eclogite Xenoliths from the Jericho and Muskox Kimberlites

Paleoproterozoic eclogite xenoliths from the Jericho and Muskox kimberlites located in the northern Slave craton, Canada, display geochemical and isotopic evidence of a recycled oceanic lithosphere origin. The eclogites comprise a diverse suite of kyaniteand diamond-bearing and bimineralic xenoliths that have a wide range of geochemical compositions. Calculated whole-rock eclogite compositions are broadly basaltic to picritic, ranging from 5 to 13 wt % MgO. Calculated whole-rock trace-element compositions for the eclogites show three overall REE patterns: the first pattern resembles that of oceanic gabbros with prominent positive Eu anomalies and flat heavy REE; the second pattern is similar to the first but displays marked light REE depletion that is interpreted to record partial melt loss; the third pattern lacks Eu anomalies and instead displays both gradually increasing REE contents from La to Lu similar to boninitic melts and ‘humped’ middle REE patterns similar to those observed in eclogites formed by reaction with subduction fluids. Several eclogites have garnet and/or clinopyroxene dO values 46ø, which coupled with the trace-element systematics of all the eclogites is interpreted to reflect derivation from recycled oceanic lithosphere protoliths. Lead isotope data for eclogite clinopyroxenes form a broad array that extends from highly radiogenic isotope compositions similar to the host Jericho kimberlite to an intersection with the Stacey^Kramers terrestrial two-stage Pb evolution curve at c. 2·2 Ga. The slope of this data array defines a secondary Pb isochron with an associated age of 1·7 0·3 Ga. Although it is possible that the Pb isotope array is actually a mixing line between the eclogite protolith and the host Jericho kimberlite, both the secondary isochron age and the c. 2·2 Ga Stacey^Kramers model age are consistent with Paleoproterozoic protolith oceanic crust formation. Eclogite xenoliths and eclogitic diamond inclusions from the Slave craton almost exclusively yield c. 2 Ga ages, which are broadly compatible with formation during subduction associated with the c. 1·9^1·8 Ga Wopmay orogeny at the western margin of the Slave craton. Trends in the Sr^Pb isotopic compositions of the eclogite clinopyroxene show that seawater alteration of the oceanic crust protoliths was an important process during protolith formation, and cryptic metasomatism further increased the Pb and Sr isotope values.This metasomatism is probably associated with theJurassic kimberlite magmatism; however, contributions from the Mesoproterozoic Mackenzie event cannot be completely excluded.

Paleoproterozoic eclogite xenoliths from the Jericho and Muskox kimberlites located in the northern Slave craton, Canada, display geochemical and isotopic evidence of a recycled oceanic lithosphere origin.The eclogites comprise a diverse suite of kyanite-and diamond-bearing and bimineralic xenoliths that have a wide range of geochemical compositions.Calculated whole-rock eclogite compositions are broadly basaltic to picritic, ranging from 5 to 13 wt % MgO.Calculated whole-rock trace-element compositions for the eclogites show three overall REE patterns: the first pattern resembles that of oceanic gabbros with prominent positive Eu anomalies and flat heavy REE; the second pattern is similar to the first but displays marked light REE depletion that is interpreted to record partial melt loss; the third pattern lacks Eu anomalies and instead displays both gradually increasing REE contents from La to Lu similar to boninitic melts and 'humped' middle REE patterns similar to those observed in eclogites formed by reaction with subduction fluids.Several eclogites have garnet and/or clinopyroxene d 18 O values 46ø, which coupled with the trace-element systematics of all the eclogites is interpreted to reflect derivation from recycled oceanic lithosphere protoliths.Lead isotope data for eclogite clinopyroxenes form a broad array that extends from highly radiogenic isotope compositions similar to the host Jericho kimberlite to an intersection with the Stacey^Kramers terrestrial two-stage Pb evolution curve at c. 2•2 Ga.The slope of this data array defines a secondary Pb isochron with an associated age of 1•7 AE 0•3 Ga.Although it is possible that the Pb isotope array is actually a mixing line between the eclogite protolith and the host Jericho kimberlite, both the secondary isochron age and the c. 2•2 Ga Stacey^Kramers model age are consistent with Paleoproterozoic protolith oceanic crust formation.Eclogite xenoliths and eclogitic diamond inclusions from the Slave craton almost exclusively yield c. 2 Ga ages, which are broadly compatible with formation during subduction associated with the c. 1•9^1•8 Ga Wopmay orogeny at the western margin of the Slave craton.Trends in the Sr^Pb isotopic compositions of the eclogite clinopyroxene show that seawater alteration of the oceanic crust protoliths was an important process during protolith formation, and cryptic metasomatism further increased the Pb and Sr isotope values.This metasomatism is probably associated with the Jurassic kimberlite magmatism; however, contributions from the Mesoproterozoic Mackenzie event cannot be completely excluded.

I N T RO D UC T I O N
The preservation of cratons since the Archean has largely been attributed to the presence of mantle roots that underlie the cratonic crust.Whether these mantle roots have remained coupled to the overlying cratonic crust since their inception is debated (e.g.Aulbach, 2012) but increasing recognition of the commonality of ages determined for the cratonic mantle lithosphere and the overlying crust suggest a shared history (Carlson et al., 2005).Major peaks in crustal growth (or reworking; Dhuime et al., 2012) in the Mesoarchean, Neoarchean, and the Paleoproterozoic are also observed in the age populations of mantle xenoliths and diamond inclusions (e.g.Pearson & Wittig, 2008;Shirey & Richardson, 2011), indicating that the cratonic crust and mantle lithosphere were at least partially coupled during growth and modification (e.g.Pearson, 1999).However, the exact mechanism of cratonic mantle lithosphere formation is highly controversial and debate continues on whether subduction-driven stacking of oceanic lithosphere (e.g.Helmstaedt & Schulze, 1989;Stachel et al., 1998;Pearson & Wittig, 2008), vertical subcretion of mantle plumes (e.g.Griffin et al., 2003;Aulbach, 2012), or a combination of the two processes was involved (e.g.Herzberg & Rudnick, 2012).Some of the arguments used in these petrogenetic models concentrate on the composition and age of kimberlite-hosted peridotite xenoliths; however, mantle eclogite xenoliths can provide equally important constraints because of their potential involvement in subduction cycles (e.g.Ringwood & Green, 1966), formation of mantle heterogeneities (e.g.Hirschmann & Stolper, 1996) and contribution to mantle melt generation (e.g.Kogiso et al., 2004).
Similar to the debate surrounding the formation of cratonic mantle lithosphere, the origin of eclogite xenoliths is also controversial.Mantle eclogite xenoliths have been interpreted as both high-pressure cumulates of mantlederived melts (e.g.O'Hara & Yoder, 1967;Smyth et al., 1989) and remnants of subducted oceanic crust (e.g.Helmstaedt & Doig, 1975;MacGregor & Manton, 1986;Jacob et al., 1994).Recent investigations of eclogite xenoliths and eclogitic diamond inclusions largely favor a recycled oceanic crust origin owing to the presence of hallmark geochemical signatures (Eu anomalies, non-'mantle' d 18 O signatures) in some samples (see review by Jacob, 2004) that are interpreted to indicate plagioclasebearing, seawater-altered oceanic basaltic and gabbroic protoliths (e.g.Schulze et al., 2003;Spetsius et al., 2009;Smit et al., 2013).However, on the basis of proposed correlations between the oxygen isotope compositions of eclogitic minerals and geochemical indices of metasomatism (Williams et al., 2009;Gre¤ au et al., 2011), some recent studies have advocated deep magmatic processes and metasomatism as critical to eclogite formation (e.g.Huang et al., 2012).
Ages determined for eclogite xenoliths and eclogitic diamond inclusions range from Neoarchean to Proterozoic, such that unraveling the petrogenesis of mantle eclogites will improve our understanding of the temporal evolution of cratonic mantle lithosphere and crust^mantle dynamics.Peridotite xenoliths and peridotitic diamond inclusions commonly yield ages older than eclogitic mantle material (e.g.Richardson et al., 1993;Aulbach et al., 2009a), and have been related to the initiation of cratonic mantle lithosphere formation (Shirey et al., 2004).However, it is significant that ages determined for cratonic mantle eclogites are largely younger than 3 Ga, suggesting that the process(es) responsible for eclogite formation and/or preservation became important only in the late Meso-to Neoarchean (e.g.Shirey & Richardson, 2011).Furthermore, ages determined from worldwide occurrences of eclogite xenoliths and eclogitic diamond inclusions largely fall into two time periods, Meso-to Neoarchean (c.3•0^2•5 Ga) or Paleoproterozoic (c.1•8^2•2 Ga), with a subordinate Mesoproterozoic (c.1^1•3 Ga) population [see compilation by Shirey & Richardson (2011)].Significantly, these eras mark prominent episodes of worldwide continental crust formation (e.g.Hawkesworth & Kemp, 2006), amalgamation (e.g.Bleeker, 2003) and modification (e.g.Griffin et al., 2004a), and the evidence of these events is preserved in cratons today.
Our study aims to further investigate the links between cratonic crust and mantle lithosphere through the study of a compositionally diverse suite of eclogite xenoliths from the Jericho and Muskox kimberlites of the northern Slave craton, Canada.Previous investigations of the eclogite xenoliths from the Jericho kimberlite have shown wide mineralogical and compositional ranges (Kopylova et al., 1999a), but significant attention has been given to two unusual eclogite types: high-FeO zircon-bearing eclogites and high-MgO diamond-bearing eclogites.Detailed geochemical investigation of the high-MgO diamond-bearing eclogites revealed a complex origin (De Stefano et al., 2009;Smart et al., 2009Smart et al., , 2012)), but isotopic investigations did not provide age constraints (Smart et al., 2009).In contrast, the high-Fe zircon-bearing eclogite xenoliths yield c. 1•9 Ga formation ages, and are interpreted as remnants of oceanic crust subducted beneath the northern Slave craton during the Paleoproterozoic Wopmay orogeny (Schmidberger et al., 2005).Despite detailed investigations of the more atypical Jericho eclogites, an integrated geochemical and isotopic study is lacking for the majority of the Jericho eclogites.To better understand the petrogenesis of the Jericho eclogite suite as a whole, and to investigate any relationship between their origin and processes occurring in the Slave cratonic crust and mantle, we have conducted a more comprehensive major-and trace-element, and Sr^Pb^O isotopic study of a wide range of eclogite types from the Jericho kimberlite, and two additional diamondbearing eclogites from the nearby Muskox kimberlite.

G E O L O GY O F T H E S L AV E C R ATO N
The Archean Slave craton of NW Canada (Fig. 1) comprises an ancient (4•03^2•85 Ga) basement complex in the central and western parts of the craton and a juvenile, 2•72^2•55 Ga volcanic and granitoid terrane in the east (Bleeker & Hall, 2007).The two terranes were amalgamated at c. 2•7^2•6 Ga and stitched together by intrusion of 2•60^2•58 Ga granites (Bleeker et al., 1999;Davis et al., 2003).The Slave craton was modified in the Paleoproterozoic at its southern and eastern margins by the TaltsonT helon orogeny, and at the western margin by the Wopmay orogeny (Hoffman, 1989).The Wopmay orogeny was accompanied by subduction of oceanic crust beneath the craton, which is hypothesized to be responsible for extensive eclogite and diamond formation in the cratonic mantle (Schmidberger et al., 2005(Schmidberger et al., , 2007;;Aulbach et al., 2009aAulbach et al., , 2009b)).This paleo-subduction zone has been geophysically imaged dipping eastward below the craton (Cook et al., 1999).Abundant mafic dykes intruded the craton between 2•2 and 1•8 Ga (Le Cheminant & van Breemen, 1994), and again at 1•27 Ga as part of the Mackenzie Igneous Event, which radiated from the NW of the craton (Le Cheminant & Heaman, 1989).Phanerozoic kimberlites occur throughout the Slave craton and are particularly concentrated in the Lac de Gras area in the center of the craton (Creaser et al., 2004).
The lithospheric mantle underlying the Slave craton is vertically and horizontally stratified with respect to its chemical composition (e.g.Griffin et al., 1999).Based on the CaO^Cr 2 O 3 compositions of garnet xenocrysts from Slave kimberlites, Gru« tter et al. (1999) divided the Slave cratonic mantle into NE^SW-trending northern, central and southern domains.The Jurassic-age Jericho (173•1 AE1•3 Ma, Heaman et al., 2006) and Muskox kimberlites lie in the northern domain (Fig. 1), and in contrast to the central and southern domains, which contain a shallow ultradepleted harzburgitic and dunitic layer, the mantle lithosphere of the northern domain is composed of more fertile lherzolites with a significant metasomatic signature in the deeper cratonic mantle lithosphere (Griffin et al., 2004b).

Previous investigations of Jericho eclogites
Eclogite xenoliths from the Jericho kimberlite have been the subject of several studies in the last 15 years.The Jericho xenolith population contains a high proportion of eclogite (25%) and pyroxenite or wehrlite (16%) xenoliths, in which the eclogites record a wide range of equilibration temperatures from 600 to 11508C [calculated at 5 GPa using the calibration of Ellis & Green (1979)], whereas the pyroxenitic xenoliths record exclusively high temperatures of 1100^12008C (Kopylova et al., 1999a(Kopylova et al., , 1999b)).Integrated textural and major-element classification [using the geochemical parameters of Coleman et al. (1965)] of 41 Jericho eclogite xenoliths found that eclogites with massive textures fall into Groups A and B and those with anisotropic textures into Groups B and C (Kopylova et al., 1999a).The Jericho kimberlite also contains two distinct eclogite suites that have been intensively studied; these include high-FeO, zircon-bearing eclogites (Schmidberger et al., 2005;Heaman et al., 2006) and high-MgO, diamond-bearing eclogites (Heaman et al., 2006;Smart et al., 2009Smart et al., , 2012)).The Jericho zircon-bearing eclogites contain exceptionally Fe-rich garnet (up to 27•5 wt % FeO) and Na-rich clinopyroxene (up to 9•33 wt % Na 2 O) and classify as Group C (Heaman et al., 2006).Zircons yield U^Pb ages ranging from 2•0 to 0•8 Ga, and Lu^Hf ages of 2•1 and 2•3 AE 0•1Ga (Schmidberger et al., 2005), whereas whole-rock Nd model ages determined for select zircon-bearing eclogites range from 1 to 1•3 Ga (Heaman et al., 2006).From the age information, the Jericho zircon eclogites are interpreted as remnants of subducted oceanic crust associated with accretion of c. 1•9 Ga magmatic arcs (e.g.Hottah or Great Bear) to the western margin of the Slave craton occurring during the Wopmay orogeny (Schmidberger et al., 2005), and underwent later modification by metasomatism related to the 1•27 Ga Mackenzie Igneous Event (Heaman et al., 2006).
In contrast, the Jericho high-MgO diamond-bearing eclogites are exclusively Group A, containing pyrope-rich garnet (up to 21•2 wt %) and jadeite-poor, CaO-rich clinopyroxene (Heaman et al., 2006;Smart et al., 2009).The Jericho high-MgO diamond-bearing eclogites have unique compositions compared with diamond-bearing eclogites worldwide (Heaman et al., 2006); the high concentrations of incompatible trace elements and radiogenic Sr^Pb isotope compositions of these xenoliths suggest a complex, multi-stage petrogenesis (Smart et al., 2009).Garnet and clinopyroxene inclusions in diamond derived from the high-MgO eclogites have lower MgO and higher Na 2 O contents than the host eclogite garnet and clinopyroxene, indicating that post-diamond formation secondary processes were critical to the petrogenesis of this eclogite suite (Smart et al., 2009(Smart et al., , 2012)).Diamonds from the high-MgO eclogites have the lowest diamond d 13 C values reported to date (^41ø; De Stefano et al., 2009).Smart et al. (2011) proposed that the carbon source for these highly 13 C-depleted diamonds was derived from ancient subducted organic matter.In this study, we have investigated 23 eclogites and one pyroxenite (Jer 5) from the Jericho kimberlite, two additional diamond-bearing eclogites from the nearby Muskox kimberlite ($15 km SW of Jericho; Hayman & Cas, 2011), and garnet separates from a previously investigated zircon-bearing Jericho eclogite (MX8A; Schmidberger et al., 2005).Some of the eclogite data were presented as a range by Smart et al. (2009) for comparative purposes, but the single xenolith data were not presented.Six of the xenoliths presented here were previously studied for their major-element compositions by Kopylova et al. (1999aKopylova et al. ( , 2004)).

Eclogite xenolith descriptions
A summary of the mineralogy and texture of all the eclogites in this study and from Smart et al. (2009Smart et al. ( , 2011Smart et al. ( , 2012) is given in Table 1.Garnet is orange, red^orange, red^pink and light lilac^pink.Clinopyroxene is dark to bright medium green and, in kyanite-bearing eclogites, pale green.Most of the eclogites are bimineralic, but some contain trace amounts of red^brown rutile and glassy apatite, which are often found as inclusions within garnet.Bright to light blue, oriented kyanite grains are found in six eclogites, and light blue needles of corundum are present in one eclogite.One pyroxenite (Jer 5; 85 modal % pyroxene content) contains minor (55 modal %) orthopyroxene.Secondary fine veins with phlogopite, calcite and apatite are common.The eclogites range from very fresh with transparent clinopyroxene and garnet with clear grain boundaries, to more altered, with irregularly shaped garnet and milky or cloudy clinopyroxene in which only small 'islands' of fresh clinopyroxene remain.The eclogites have both massive, interlocking textures (generally found in bimineralic eclogites) and anisotropic, oriented textures (generally found in kyanite-bearing eclogites).Type I and II textural eclogite types, as described from South African eclogites by MacGregor & Carter (1970), are not apparent in the eclogites.Whereas major-element compositions were calculated for all eclogites (see Table 1), trace-element and isotope compositions were determined only for eclogites deemed to contain minimally altered clinopyroxene.Smart et al. (2009) reported on a large (n ¼13) suite of high-MgO diamond-bearing eclogites from Jericho; here we include the chemical composition of two additional diamond-bearing eclogites (Jde15, Jde25) which have markedly lower MgO contents and were first reported by Smart et al. (2011).In addition, we also present data for two diamond-bearing eclogites from the nearby Muskox kimberlite (Mox001, Mox023).The Jericho diamondbearing eclogites reported here contain 5^10 modal % diamond, light green, milky clinopyroxene and orange garnet and show high degrees of alteration, in contrast to the very fresh high-MgO Jericho diamond-bearing eclogites from Smart et al. (2009).The two diamond-bearing eclogites from the Muskox kimberlite contain only trace amounts of diamond and have deep red garnet and dark green clinopyroxene.

A NA LY T I C A L T E C H N I Q U E S Major-and minor-element analysis
Garnet and clinopyroxene from all xenoliths were analyzed for major and minor elements using a JEOL 8900 electron microprobe at the University of Alberta with a 20 nA beam current and 20 kV accelerating voltage.Effort was taken to analyze touching garnet and clinopyroxene pairs in core-to-rim mineral grain transects; however, for some of the diamond-bearing eclogites (Jde15 and Jde25), mineral separates mounted in epoxy mounts were used instead.Three to five grains of garnet and clinopyroxene with 10 analyses per mineral grain were obtained for each xenolith.

Trace-element analysis
Garnet and clinopyroxene were analyzed for their traceelement compositions in both grain mounts and 'thick'sections by laser ablation quadrupole inductively coupled plasma mass spectrometry (ICP-MS) at the University of Alberta using a New Wave Research Nd:YAG UP213 laser system coupled to a Perkin Elmer Elan 6000 Quadrupole ICP-MS system.A spot size of 160 mm and a fluence of c. 15 J cm À3 were used for all analyses.Background counting time was 20 s followed by 50 s of ablation.Each analytical session was bracketed by three to four analyses of the NIST 612 glass standard to monitor accuracy.Repeated analyses of the NIST 612 standard over four analytical sessions yielded relative standard deviations generally 55% (excluding 59% for Ni and Ti) and differences from the accepted values of 53 % for the trace elements of interest (excluding one session where the per cent differences were 56%).The 44 Ca content of garnet and clinopyroxene (from electron microprobe analysis) was used for calibration.All data were reduced using GLITTER data reduction software (van Achterbergh et al., 2001).The uncertainty for most elements is c. 10% relative (2s), but for elements present in very low abundance [e.g.U, Th, Rb; heavy rare earth elements (HREE) in clinopyroxene; light REE (LREE) in garnet] the uncertainty can be up to 40%.Between five and 40 grains of garnet and clinopyroxene were analyzed per xenolith with generally one spot per grain.

Sr and Pb isotope analysis
The Sr and Pb isotopic compositions of clinopyroxene separated from the Jericho xenoliths were determined using a NuPlasma multi-collector (MC)-ICP-MS system coupled to the New Wave laser system described above.When possible, Sr and Pb isotope analyses were determined on the same grain for which trace-element contents were previously analyzed.The Jericho and Muskox

Oxygen isotope analysis
The oxygen isotope composition of garnet and clinopyroxene separates was obtained by laser fluorination at the University of New Mexico, following the procedure of Sharp (1990).Inclusion-and crack-free garnet and clinopyroxene grains (1^2 mg) were loaded into a Ni block, dried in an oven and then transferred into the sample chamber to be placed under vacuum.Samples were reacted with BrF 5 and heated with a Merchantek 25 W CO 2 laser.The extracted O 2 gas sample was purified over liquid N and KCl traps before being measured on-line using a dual-inlet Finnigan MAT Delta XL mass spectrometer.Mineral standards were measured throughout the analytical session (Gee Whiz quartz Valley et al., 1995) and were within error of the accepted values (measured Gee Whiz quartz:

M I N E R A L C H E M I ST RY
The major-and trace-element compositions of garnet and clinopyroxene from the Jericho eclogites are listed in Tables 2 and 3, respectively.

Major and minor elements
All garnets from the Jericho and Muskox eclogites studied here are Cr-poor (50•13 wt %) pyrope^grossular^almandine solid solutions, and broadly overlap in composition with garnet from the Jericho eclogites investigated by Kopylova et al. (1999a) andDe Stefano et al. (2009).
Garnet has variable FeO (11•6^26•4 wt %), MgO (5•0^15•3 wt %) and CaO (4•0^17•7 wt %) contents with Mg-numbers ranging from 27 to 71.The Ca^Mg^Fe compositions of garnet from the present study are plotted in Fig. 2 and classify as Group 'B' and 'C' using the geochemical classification of Coleman et al. (1965).The new garnet compositions overlap with those from previous studies of Jericho and Slave eclogites (Fig. 2).Group C garnets on average have higher CaO contents than group B garnets (mean CaO of 13•1 vs 8•0 wt %), which are more Mg-rich (Mg number 57 vs 39).Garnet from the diamond-bearing Jericho and Muskox eclogites studied here (white diamond symbols in Fig. 2) is more pyropic than the non-diamondbearing eclogites, falling within the Group B field (Fig. 2), and has uniform Mg-numbers (66^67) with a much smaller range in CaO contents (6•8^9•1wt %).It should be noted that the Group B diamond-bearing eclogites reported here have much lower MgO contents compared with the Jericho high-MgO diamond-bearing eclogites (Group A in Fig. 2) from Smart et al. (2009).
Clinopyroxenes from the Jericho eclogites also have a wide range of compositions, and overlap with clinopyroxenes from previous Jericho eclogite studies (Fig. 3).The Jericho clinopyroxenes classify as Group B and C using the MgO vs Na 2 O diagram from Taylor & Neal (1989) (Fig. 3), in agreement with the garnet classification in Fig. 2. Clinopyroxenes from Group C eclogites have on average higher Na 2 O (7•3 vs 5•0 wt %; Fig. 3a) and Al 2 O 3 (13•3 vs 8•9 wt %) contents than the Group B clinopyroxene, but also have lower CaO contents (12•1 vs 15•6 wt %; Fig. 3b) and Mg-numbers (75 vs 83).The diamond-bearing eclogite clinopyroxenes have more jadeite-poor and diopside-rich compositions compared with clinopyroxene from the diamond-free eclogites, and classify as Group B (Fig. 3a).Of note, the Group B diamond eclogites reported here have significantly lower MgO and CaO, but higher Na 2 O contents than the clinopyroxene from the high-MgO diamond-bearing eclogites from Jericho.

Trace elements Garnet
A subset of the 26 Jericho eclogites analyzed for major elements was selected for trace-element analysis based on degree of freshness and preservation of fresh clinopyroxene.Grain mounts were made from mineral separates for ten of the eclogites (eclogites 35-2 to JD6fn; Mox001; Mox023 in Table 3), whereas thick-sections were made of the remaining ten eclogites (Jer 1 to Jer 8; Jde15; Jde25) owing to their small size.
In contrast to garnet from the Group B and C eclogites, garnets from the Jericho and Muskox Group B diamondbearing eclogites (Fig. 4c) have flat to positively sloping middle REE (MREE)^HREE patterns (Lu N /Gd N Jericho ¼ 0•4^1; Muskox ¼ 2•5^4•7).Eclogite Jde25 has higher amounts of MREE that form an elevated plateau from Eu to Dy and result in a low Lu N /Gd N value of 0•4.However, the diamond-bearing eclogite garnets have lower HREE contents (Lu N ¼ 4•9^12) compared with garnets from the diamond-free eclogites (Lu N up to 34), and Eu anomalies are small or absent (Eu/Eu* ¼ 0•9^1•3).Some garnets also show enrichment in La N compared with the other LREE, and these garnets also have high Ba contents (0•81^1•53 ppm), which may be attributed to accidental analysis of infiltrated kimberlite material during laser ablation.Ni contents in garnet from the diamond-bearing eclogites range from 22 to 67 ppm.

Clinopyroxene
The REE patterns for clinopyroxene from the Jericho and Muskox eclogites are shown in Fig. 4d for Group B, 4e for Group C and 4f for the diamond-bearing eclogites.These Group A^B^C divisions are from Taylor & Neal (1989).Clinopyroxenes from the diamond-bearing eclogites have relatively flat, slightly negatively sloping LREE patterns (Fig. 4f).LREE contents in these clinopyroxenes fall at the lower end of the range observed in diamond-free Jericho eclogites (La N ¼ 0•7^4•4 vs up to 18).Positive Sr anomalies in primitive mantle-normalized multi-element patterns are also observed in clinopyroxenes from diamond-bearing eclogites.However, both Sr (56^83 ppm) and Pb (0•1^0•2 ppm) contents are much lower in the diamond-bearing eclogites than in the diamond-free eclogites.

G E OT H E R M O M E T RY
Temperatures were calculated at 5 GPa using the Krogh Ravna (2000) calibration of the garnet^clinopyroxene Fe^Mg exchange thermometer.Because no geobarometer exists for bimineralic eclogites, we used our calculated temperatures to extrapolate pressures from the pressuret emperature array defined for Jericho peridotite xenoliths (Kopylova etal.,1999b).Pressures andtemperatures were successively iterated using this technique and are given inTable 2. Temperatures for the diamond-free eclogites range between 840 and 14108C, corresponding to a range in estimated pressures from 4•1to 6•0 GPa.The choice of an initial pressure of 5 GPa (corresponding approximately to1601 70 km depth) maybe excessive for some of the eclogites; however, conducting the iteration with an initial lower pressure (e.g. 4 GPa) produces the same final temperatures and pressures after several iterations.Calculated temperatures for the diamond-bearing eclogites are in the range of 10701 1008C (Jericho) and1120^12208C (Muskox), corresponding to pressures of 4•6^5•0 GPa, and place the diamond-bearing eclogites at the higher end of the temperature range defined by the diamond-free Jericho eclogites.Diamonds from two Jericho eclogites (Jde15, Jde25) have independent temperature estimates from nitrogen aggregation systematics.Assuming diamond formation at 2 Ga synchronous with both Jericho eclogite formation (Schmidberger et al., 2005) and Diavik diamond formation (Aulbach et al., 2009a), two diamonds from Jde15 (eclogite temperature of 10401 0808C), containing1130^2350 ppm Nand with low aggregation (6^10 %B, where %B indicates the proportion of nitrogen aggregated to diamond B centers), yield nitrogen aggregation temperature estimates of 1040^10508C.Ten diamonds fromJde25 (eclogite temperature 1040^11008C), containing 1080^1990 ppm N with variable aggregation (12^56 %B) give temperature estimates between 1040 and 11008C.Thus the nitrogen aggregation diamond-derived temperatures (1040^11008C) agree with the Fe^Mg exchange eclogite-derived temperatures (1070^11108C) within the uncertainty of the two techniques.
TwoJericho eclogites give extreme temperatures414008C and appear to have equilibrated at conditions deeper than calculated for the magmatic pyroxenites and fertile, porphyroclastic high-T peridotites that are thought to define the base of the Jericho cratonic mantle lithosphere (Kopylova et al.,1999b).These geologically unreasonable depths may indicate that the minerals analyzed in these eclogites were not in equilibrium.Alternatively, temperature overestimation could have resulted from using total FeO in the temperature calculation, which, when applied to low-Fe minerals (such as eclogitic omphacite), can produce misleadingly high temperatures (Krogh Ravna, 2000).

C L I N O P Y ROX E N E S r A N D P b I S O TO P E C O M P O S I T I O N S
87 Sr/ 86 Sr ratios were determined for clinopyroxene from 15 eclogites (averages and calculated uncertainties for each eclogite are listed in Table 4; full data are given in Supplementary Data Appendix A; supplementary material is available for downloading at http://www.petrology.oxfordjournals.org)and have a range of 0•7032^0•7054 and 131^678 ppm Sr.The Sr isotope compositions reported in Table 4 have been corrected for radiogenic ingrowth of 87 Sr since 173 Ma, but because most of the analyzed clinopyroxene grains have very low contents of Rb (below detection to 0•45 ppm), the corrections were very small (50•0001 change).Whole-rock analyses of the Jericho kimberlite have a range of 87 Sr/ 86 Sr ratios from 0•7043 to 0•7063 (Kopylova et al., 2009), which overlap with those of the Jericho Group A diamond-bearing eclogites (0•70470 •7061, Smart et al., 2009).However, 11 of the 15 Jericho eclogites (this study) have clinopyroxene with 87 Sr/ 86 Sr ratios between 0•7036 and 0•7044 and no correlation between Sr isotope ratios and Sr content is apparent (r 2 ¼ 0•3, not shown).
Lead isotope ratios ( 206 Pb/ 204 Pb and 207 Pb/ 204 Pb) were determined for clinopyroxene from 13 eclogites (averages and uncertainties for each eclogite are listed in Table 4; full data are given in Supplementary Data Appendix A) and show a range of 14•68^18•74 and 15•02^15•54, respectively.The Pb isotope data presented here have not been corrected for isotopic ingrowth because 238 U/ 204 Pb ('m') values were not measured.However, we assume that our measured Pb isotope values approximate the initial Pb isotope values at the time of eclogite formation (using the 'galena model age' technique) given that the clinopyroxene grains have both very low U contents (below detection to 0•03 ppm) and very low U/Pb ratios (0•01^0•09).However, we stress that the Pb isotope values presented here are maximum values and correction to initial values may shift the isotopic array to less radiogenic values.
In Fig. 5, the Jericho data plot along a broad array from an intercept with the two-stage terrestrial lead evolution curve of Stacey & Kramers (1975)

OX YG E N I S O T O P E C O M P O S I T I O N S
The oxygen isotope compositions of garnet, clinopyroxene and kyanite mineral separates from 10 of the Jericho eclogites are presented in  Mantle xenoliths are commonly infiltrated by host kimberlite magmatic fluids or melts, such that bulk xenolith analyses yield contaminated, unrepresentative compositions (e.g.Barth et al., 2001;Schmidberger & Francis, 2001;Jacob, 2004;Aulbach et al., 2007).Therefore, the wholerock compositions of eclogite xenoliths are generally estimated using modal abundance data and the composition of constituent minerals.However, inaccurate mode estimates owing to small xenolith size, large mineral grains or mineral nugget effects can produce erroneous calculated compositions.For example, coarser-grained (mineral grains exceeding 5 mm) samples are thought to have AE10% uncertainties in modal estimations (Jerde et al., 1993).Calculated major-element whole-rock compositions are particularly mode-dependent (e.g.Smart et al., 2009), which must be taken into consideration when using calculated major-element whole-rock compositions.In contrast, the overall shape of whole-rock trace-element patterns are relatively unaffected by up to 30% variation in mineral modes (Jerde et al., 1993).We have calculated both majorand trace-element whole-rock compositions for the Jericho and Muskox eclogites using both visually estimated and point-counted (using thin sections) modal abundances.Whereas some of the Jericho eclogite samples are small (5 cm in length), the mineral grain diameters are also relatively small (1^3 mm), such that we are confident our estimates are within 10^20% of the actual mode.One exception is sample 44-9, which contains coarser crystals up to 5 mm in diameter.Calculated major-element wholerock compositions are listed in Table 2 and trace-element compositions in Table 6.

Trace-element classification
The calculated whole-rock trace-element compositions of the Jericho and Muskox eclogites are listed in Table 6 and plotted in Figs 7 and 8.The eclogites of this study have been grouped into three REE pattern types and the averages for each pattern type are shown in Fig. 7a.The first pattern (Fig. 7b) is characterized by slightly positively On primitive-mantle normalized multi-element plots, the three trace-element pattern types also have distinguishing features (Fig. 8a).The peaked and depleted eclogites have generally similar patterns, but as observed in Fig. 7, depleted eclogites exhibit stronger LREE depletion and HREE enrichment than the peaked eclogites (Fig. 8b  and c).The sloped eclogites have lower REE abundances, but moderately higher abundances of some of the large ion lithophile elements (LILE; Rb and Ba) and high field strength elements (HFSE; Th, U, Nb, Ta, Zr, Hf and Ti) compared with the other two patterns (Fig. 8d).Strong positive Sr anomalies are observed in all three eclogite patterns, but depletions in the HFSE, observed as sharp negative anomalies in both Ti and Nb, and troughs at Zr^Hf, are only found in the peaked and depleted eclogites.Some HFSE depletion may be due to the omission of rutile from the calculated bulk-rock eclogite, as rutile, although present in some eclogites in trace to minor amounts, was not analyzed for its trace-element content.After incorporation of 0•05^2% rutile into the peaked and depleted wholerock eclogite compositions [using data from Aulbach et al. (2011a)], the Ti and Hf depletions are erased, but Zr depletions are still prominent (not shown).BothTa and Nb concentrations are increased slightly (not shown), but the relative Nb depletion and Ta peak remain in the eclogite pattern.Addition of minor rutile indicates that some of the HFSE anomalies, such as Ti and Hf, may be an artefact of the omission of rutile in the bulk-rock calculation.
This new reconstructed bulk-rock trace-element grouping of the Jericho and Muskox eclogites does not coincide with the garnet-based major-element classification discussed above; rather both Group B and C Jericho eclogites fall into the peaked and depleted groups.However, the peaked eclogites are all Group B except for one Group C eclogite (10-13) and the diamond-bearing eclogites exclusively make up the sloped group.

D I S C U S S I O N The effect of metasomatism
Mantle eclogites are particularly susceptible to metasomatism because of their protracted residence in the cratonic lithospheric mantle and subsequent entrainment as xenoliths in kimberlite magmas.Metasomatism is generally expressed as either the growth of new minerals (modal 1s calculated for each day session based on the Gee Whiz Qtz standard.Gore-3 is sampled from the same locality as UWG-2.*Diamond-bearing samples. metasomatism) and/or trace-element enrichment without new mineral growth (cryptic metasomatism; Dawson, 1984).It is critical to evaluate the potential effects of metasomatism in xenolith studies so that the petrogenetic history of the pre-metasomatic protolith may be determined.
The Jericho and Muskox eclogites of this study do not show strong evidence of modal metasomatism, and although secondary veins of fine-grained phlogopite, calcite and apatite are common, these are regarded as products of kimberlite infiltration.In comparison, diamond found in the Jericho and Muskox eclogites is clearly a secondary mineral and it is often surrounded by altered clinopyroxene and garnet.It is possible that diamond growth was accompanied by cryptic metasomatism in these eclogites; however, comparison of the diamond-bearing eclogites of this study and the high-MgO diamond-bearing eclogites from Smart et al. (2009) in Figs 7d and 8d demonstrates that the diamond-bearing eclogites reported here do not display the extreme LREE enrichment found in the high-MgO diamond-bearing eclogites, and thus have not undergone significant metasomatic enrichment.Therefore, excluding the diamond-bearing eclogites, clear evidence for modal metasomatism is absent in the Jericho eclogites.
Comparison of incompatible element contents in minerals from eclogites with clear evidence of modal metasomatism and the Jericho eclogites of this study (Fig. 9) demonstrates that garnet and clinopyroxene from modally metasomatized eclogites are clearly more enriched in incompatible elements (represented by Zr and Ce).In Fig. 10, we compare the eclogites of this study with the metasomatized zircon-bearing high-FeO and diamond-bearing high-MgO eclogites from the Jericho kimberlite (Heaman et al., 2002;Smart et al., 2009).Both the high-MgO diamond-bearing and zircon-bearing eclogite suites have higher concentrations of REE from La to Sm, whereas the zircon-bearing eclogites have much higher concentrations of the LILE and HFSE, up to 1000 times primitive mantle.It should be noted that the data for the zircon-bearing eclogites from Heaman et al. (2002) are from bulk-rock eclogite analyses, which is probably responsible for the extreme traceelement concentrations observed in these eclogites.From this comparison, it appears the Jericho eclogites of the  Rollinson (1999), Yamashita et al. (2000) and Polat et al. (2007Polat et al. ( , 2008)).
present study have not been strongly affected by metasomatic processes and we can confidently use the eclogite bulk-rock geochemical data to assess the potential eclogite protoliths and their petrogenetic histories.

Crustal vs mantle eclogites
Kimberlite-hosted mantle-derived eclogite xenoliths commonly have basaltic to gabbroic compositions that overlap those of Archean basalts, komatiites and modern oceanic gabbros (e.g.Barth et al., 2001;Jacob, 2004).These compositional similarities between eclogites and oceanic rocks are also shown by the Jericho and Muskox eclogites of this study (Fig. 6).The broadly basaltic composition of eclogite xenoliths has caused much debate about their origin; they have been hypothesized to form either in the upper mantle as magmatic cumulates crystallized from peridotite-derived basaltic magmas (e.g.O'Hara & Yoder, 1967;Griffin & O'Reilly, 2007), or in the oceanic lithosphere as basalts or gabbros, before being recycled into the cratonic lithospheric mantle (e.g.Helmstaedt & Doig, 1975;Jagoutz et al., 1984;Jacob et al., 1994;Barth et al., 2001).However, although a bulk-rock basaltic composition alone cannot discriminate between a 'shallow' oceanic lithosphere or a 'deep' mantle magmatic origin for eclogites, their trace-element and oxygen isotope compositions can provide better constraints on their origin(s).At present there is no definitive evidence for the formation of eclogites as mantle cumulates (e.g.Jacob, 2004), but compelling evidence for a recycled oceanic crust origin is provided by the range of oxygen isotope compositions of mantle eclogites, which overlaps with both seawater-altered, modernday oceanic crust (Muehlenbachs and Clayton, 1972;Alt et al., 1989) and oceanic crust preserved in ophiolite complexes (Gregory & Taylor, 1981), and deviates significantly from the average mantle value of 5•5 AE 0•4ø (Mattey et al., 1994).Given the limited fractionation of oxygen isotopes at high temperatures at mantle conditions (Clayton et al., 1975;Eiler, 2001), any deviation of eclogite oxygen isotope compositions outside the mantle average is probably due to protolith interaction with seawater near the Earth's surface (Muehlenbachs and Clayton, 1972).Garnets from the Jericho eclogites have oxygen isotope compositions ranging from 5•3 to 6•6ø; however, only three of the nine eclogites analyzed for their oxygen isotope composition have garnet d 18 O values outside the mantle range of 5•5 AE 0•4ø.Garnets from these three samples have d 18 O values 46ø and comprise one peaked (10-13) and two sloped eclogites (Jde15, Jde25).Although most of the measured eclogites have 'mantle-like' oxygen isotope compositions, this does not preclude an oceanic crustal protolith for these eclogites (e.g.Schmickler et al., 2004;Smart et al., 2012).Oxygen isotope depth profiles through the oceanic crust show that the upper part of the sheeted dyke layer in the oceanic crustal column has d 18 O values close to 5•5ø (Alt et al., 1989), despite the fact that these rocks have undergone a significant amount of seawater alteration.The occurrence of mantle-like d 18 O values in the upper sheeted dykes reflects the fact that this depth marks the transition from lower temperature sea-water^rock interactions in the shallow pillow basalts (d 18 O46ø), to higher-temperature water^rock interactions in the lower sheeted dykes and upper gabbros (d 18 O55ø).Furthermore, some deeper sections of the oceanic crust (e.g.middle to lower gabbros) may escape seawater alteration altogether and thus preserve mantlelike oxygen isotope compositions (e.g.Gregory & Taylor, 1981;Hart et al., 1999).Despite the predominant 'mantle' signature of the oxygen isotope compositions, the Jericho eclogites have trace-element characteristics that are indicative of protolith crystallization at low pressures.These characteristics are examined in the following discussion.

Peaked and depleted eclogites
Protolith formation as oceanic gabbros The Jericho and Muskox eclogites have variable compositions ranging from more Ni-rich, picritic compositions with 13 wt % MgO, to more Al 2 O 3 -and Na 2 O-rich, basaltic compositions with 5•3 wt % MgO and low Ni contents (Fig. 6a, e and f).The trace-element patterns defined in this study for the Jericho eclogites broadly correspond to the major-element groups.Specifically, the peaked (and sloped) eclogites correspond to the more MgO-rich picrite group, and the depleted eclogites to the Al-rich basaltic group (Fig. 11).The well-defined trend between MgO vs Al 2 O 3 displayed by these eclogites is very similar to magmatic differentiation trends defined by both natural (Cox, 1980;Eggins, 1993;Benoit et al., 1996) and experimental (Baker & Stolper, 1994) picrites and basalts, which have been superimposed on Fig. 11.The similarity between the differentiation and eclogite trends suggests that the eclogite compositional variation may be related to sampling of oceanic rocks that underwent variable amounts of crystal fractionation.It is not possible to determine the exact mineralogy of the eclogite protoliths; nevertheless, similarities to the gabbroic cumulates in ophiolites (Ni 130^360 ppm; MgO 4•5^12•5 wt %; Benoit et al., 1996), as well as the overall low Ni contents (5 10^180 ppm) at low to moderate    MgO contents (5•3^13•7 wt %) indicate that olivine was probably not an abundant mineral phase (e.g.Jacob & Foley, 1999), and the eclogites were probably plagioclaseand clinopyroxene-rich gabbroic rocks.
One aspect that is problematic for this hypothesis is the high FeO content (4 13 wt %) found in several calculated eclogite compositions (Fig. 6c).There is a positive correlation between FeO and Y content (r 2 ¼ 0•6, not shown) and eclogites with the highest FeO (up to 21•8 wt %) and Y (up to 56 ppm) contents are also the most garnet rich (eclogites 35-2, 10-13, Jer 4, Jer 6 and Jer 8).However, adjustment of the garnet:clinopyroxene modes to 50:50 does not diminish the trends and absolute FeO contents remain high (up to 17•0 wt %).Whereas these eclogites do not overlap with the natural gabbro field shown in Fig. 6, some of the Archean basalts plotted do have higher FeO contents, up to 15 wt %.Thus the elevated FeO contents, although still enigmatic, may be explained by either derivation from a high-FeO precursor or accumulation of almandine-rich garnet during eclogitization.
The presence of Eu and Sr anomalies in the peaked and depleted eclogites could suggest that plagioclase accumulation occurred in the protoliths.Comparison of these two eclogite groups with both modern oceanic gabbros (Bach et al., 2001) and gabbroic cumulates from the Oman ophiolite (Benoit et al., 1996) in Fig. 12 demonstrates the strong similarities between these lithological groups.Both the oceanic gabbros and the Jericho peaked and depleted eclogites exhibit positive Eu, Sr and Pb anomalies and negative Zr anomalies, as well as flat HREE patterns (Figs 7 and 8).
Therefore, we suggest that the peaked and depleted Jericho eclogites represent gabbroic cumulates formed by variable fractionation of plagioclase, clinopyroxene and

Jericho zircon eclogites
Jericho high-MgO diamond eclogites Fig. 10.Comparison of the three eclogite patterns from this study (symbols as in Fig. 8) with the high-MgO diamond-bearing eclogites and zircon-bearing eclogites from the Jericho kimberlite that contain evidence of modal and cryptic metasomatism.The high-MgO diamond-bearing eclogites (Smart et al., 2009) have much higher contents of REE from La to Sm, and also of Nb, Pb, Zr and Hf compared with the Jericho eclogites of this study.Zircon-bearing eclogites (Heaman et al., 2002) have overall much higher trace-element contents than the eclogites studied here and also display extremely high HFSE contents.Primitive-mantle normalization values are from McDonough & Sun (1995).olivine from evolving basaltic^picritic magmas during formation of ancient oceanic crust.Trace-element modeling of a gabbroic cumulate crystallized from a partial melt of depleted MORB mantle (model details are described in the caption to Fig. 13) shows that the resulting REE pattern has a positive Eu anomaly of 1•8, flat HREE and a slight depletion in the LREE (Fig. 13a).The model gabbro REE pattern resembles both the peaked and depleted eclogites (Fig. 13a), but is lacking some of the more pronounced LREE depletion observed in the depleted eclogites.The REE pattern of the peaked eclogites is strikingly similar to that of modern oceanic and ophiolitic gabbros (Fig. 13a); in comparison, the depleted eclogites have lower LREE and higher HREE contents.This depletion in the most incompatible elements, in combination with the lower SiO 2 contents observed in the depleted eclogites   (1996).Picrite differentiation trends are from Eggins (1993) and Alonso-Perez (2006).Field of experimental melts of peridotite at 1GPa is from Baker & Stolper (1994).
(Fig. 6e), is evidence for removal of a partial melt from these eclogites.

Effect of melt depletion
Low SiO 2 for a given MgO content is often observed in eclogite xenoliths when compared with both oceanic crust and Phanerozoic crustal eclogites (Pearson et al., 2003), and is commonly ascribed to partial melt extraction (e.g.Ireland et al., 1994;Jacob & Foley, 1999).Partial melting of oceanic crust in Archean subduction zones has often been cited as the mechanism for tonalite^trondjhe-mite^granodiorite (TTG) production and cratonic crust growth (Ireland et al., 1994), which left behind eclogite residues that were entrained in the cratonic mantle lithosphere.Jacob & Foley (1999) Workman & Hart (2005)] using the partition coefficients of Kelemen et al. (1993).
The gabbro composition was calculated assuming 90% fractional crystallization of this melt using the partition coefficients for plagioclase, pyroxenes and olivine from Norman et al. (2005).The modal abundance of plagioclase, clinopyroxene, orthopyroxene and olivine (53%, 20%, 0%, 27%) used for the calculated gabbro was determined from the average CIPW norms calculated for the whole-rock compositions of peaked Jericho eclogites.Field of oceanic gabbros is from Bach et al. (2001) and representative ophiolite olivine gabbro is from Benoit et al. (1996).(b) Jericho depleted eclogites are modeled as residual peaked eclogites after partial melt extraction at high pressure.Incremental 5, 10 and 20% batch partial melts were extracted from the average peaked Jericho eclogite using partition coefficients for gabbroic-protolith eclogite/melt from Yaxley & Sobolev (2007) and a mode of 50% garnet and 50% clinopyroxene.Chondrite normalization values are from McDonough & Sun (1995).
'restore' the SiO 2 composition of mantle eclogite xenoliths back to basaltic compositions.Here, we apply the same principle, and show that addition of 5^10% tonalitic melt to the calculated depleted eclogite compositions is sufficient to reach the SiO 2 contents of Archean basalts or modern gabbros (dashed lines in Fig. 6e).Trace-element modeling of 5^20% bulk partial melt extraction from the peaked eclogites, shown in Fig. 13b (details of the modeling are given in the figure caption), produces the appropriate range of LREE depletion displayed by the depleted eclogites.Thus, variable amounts of partial melt extraction from the peaked group of eclogites appear to be responsible for the composition of the depleted eclogites.

Sloped eclogites
Protolith formation as second-stage melts In contrast to the gabbroic nature of the Jericho peaked and depleted eclogites, the sloped eclogites lack Eu and Zr anomalies, display both concave-up REE (Jericho) and fractionated HREE (Muskox) patterns (Fig. 7d), and have relatively smaller positive Sr anomalies (Fig. 10).
Additionally, the sloped eclogites are compositionally more uniform and appear to be more 'primitive', having elevated Mg-numbers of 75^76 and overall lower contents of REE (Figs 7a and 8a).Combined with the elevated d 18 O values of 6•5^6•6ø in these eclogites, we suggest an origin as seawater-altered basaltic^picritic lavas rather than plagioclase-rich gabbros.However, the REE patterns of the Jericho and Muskox sloped eclogites do not resemble typical normal mid-ocean ridge basalt (N-MORB) or enriched (E-)MORB patterns; instead, the overall lower trace-element contents and higher Mg-numbers imply that the magma source for these eclogites was similarly depleted.Low-Mg and high-Ca eclogites from the Lac de Gras kimberlites, 100 km south of Jericho, have traceelement patterns that resemble those of the Muskox and Jericho sloped eclogites, respectively.The low-Mg eclogites of Diavik (Aulbach et al., 2007) and the Muskox eclogites of this study have similar positively sloping HREE patterns and subchondritic Zr/Hf values, features that Aulbach et al. (2007) noted as similar to melts formed from previously depleted sources, such as boninites (e.g.Wyman, 1999;Polat et al., 2002;Smithies et al., 2004).The resemblance between the Muskox sloped eclogites and boninites is demonstrated in Fig. 14, and the higher MgO contents in these eclogites support derivation from previously depleted sources.
In contrast, the Jericho sloped eclogites with concave-up REE patterns and a strong MREE plateau (see Jde25) resemble the high-Ca and diamond-bearing HREE-enriched eclogites from the Lac de Gras area (Aulbach et al., 2007(Aulbach et al., , 2011b)).The Lac de Gras eclogite groups were proposed to form by fluid-mediated eclogitization of a subducting slab, driven by the influx of massive amounts of subduction fluids related to serpentinite dehydration in the deeper mantle sections of the slab (Aulbach et al., 2011b).Considering the similar concave REE pattern with an MREE plateau, a subduction-fluid mediated origin may also be acceptable for the Jericho sloped eclogites.Diamond formation is intimately associated with the serpentinite-fluid eclogitization in the Lac de Gras eclogites (Aulbach et al., 2011b), and an oceanic mantle lithospherederived fluid may also serve to explain the presence of diamond with broadly mantle-like d 13 C values between ^3•5 and ^5•3ø (Smart et al., 2011) in the Jericho sloped eclogites.

Isotopic composition and possible age of the Jericho eclogites Pb isotope age constraints
The geochemical data suggest that, despite some petrological differences, all three Jericho eclogite groups probably had oceanic lithosphere protoliths.In the Slave craton there are two proposed subduction events that may have resulted in the incorporation of oceanic crust into the lithospheric mantle: (1) subduction associated with juvenile terrane accretion and craton amalgamation at c. 2•7^2•6 Ga (Davis et al., 2003); (2) collision of the Hottah terrane and development of the Great Bear Magmatic arc as part of the Wopmay orogeny at c. 1•8^2•1 Ga (Hildebrand et al., 1987;Cook, 2011).If the Jericho Pb isotope data array (Fig. 5) represents a secondary isochron, then the slope of that array yields a Paleoproterozoic age of 1•7 AE 0•3 Ga, which, within error, is consistent with the age of the Wopmay orogeny, and may represent the timing of eclogitization during subduction of oceanic crust at the western margin of the Slave craton.However, another possibility is that the Pb isotope array shown in Fig. 5 represents mixing of Pb between the eclogite protolith and the host kimberlite.Isotopic mixing arrays between 'protolith' and host kimberlite are commonly used to explain isotopic variations in mantle samples (e.g.Schmidberger et al., 2007;Tappe et al., 2011;Hunt et al., 2012); however, in such cases, elemental and isotopic ratio correlations are often observed.There is no correlation of Pb content with 206 Pb/ 204 Pb ratios in the Jericho clinopyroxenes (r 2 ¼ 0•09, not shown) but it is possible that the Pb systematics of the Jericho eclogites may have been disturbed by secondary processes.
Diffusion of Pb isotopes in clinopyroxene is relatively slow even at elevated temperatures, such that original compositions can be retained by 1mm grains of clinopyroxene at moderate cratonic mantle temperatures of 9008C for 10 Myr timescales, although this has been shown to be highly dependent on clinopyroxene composition and oxygen fugacity (Cherniak, 2001).Most of the Jericho eclogites record a range of equilibration temperatures from 850 to 12508C and contain 1^3 mm clinopyroxene grains.If contact with any metasomatic agents (e.g. the host kimberlite before eruption) or exchange with high-m lithologies occurred only on short timescales, then it is possible that the clinopyroxene grains could retain their near-primary Pb isotope composition.Coupled with the low U/Pb values in the Jericho clinopyroxene and assuming the Pb isotope compositions were 'frozen in' the clinopyroxene shortly after eclogitization of the subducting slab (owing to slab melting; see below), we argue that the Pb isotope compositions of the clinopyroxene closely represent that of the protolith.
To determine a common Pb model age for the Jericho eclogites, we applied the Stacey & Kramers (1975) twostage Pb evolution model.This model is a reasonable approximation of the Pb isotopic evolution in the upper mantle (Stacey & Kramers, 1975), which in turn is an appropriate model reservoir for extraction of the Jericho oceanic lithosphere protoliths.Eclogite Jer 3 intersects the Stacey^Kramers curve at c. 2•2 Ga, and provides good evidence for Paleoproterozoic formation of the Jericho protoliths.However, it must be stressed that the 2•2 Ga model age is a minimum age.It is possible that at some point during eclogite petrogenesis, clinopyroxene could have evolved at a high-m value such that the least radiogenic isotopic composition preserved in the Jericho clinopyroxene ( 206 Pb/ 204 Pb ¼14•7) is not representative of the initial isotopic ratio as assumed above.This speculation is significant in that it would imply pre 2•2 Ga (Archean?)eclogite formation and preservation in the northern Slave craton.However, Archean eclogite formation is not considered likely as there is no independent line of evidence for older eclogite formation, and the Pb model ages determined here overlap well with Hf model ages of 2•1 AE0•1 and 2•3 AE 0•1Ga determined from zircons extracted from the Jericho high-FeO eclogites (Schmidberger et al., 2005).
The minimum model age of 2•2 Ga is $100^300 Myr older than the known subduction events associated with formation of the c. 2•1^1•9 Ga Hottah arc terrane or c. 1•87^1•84 Ga Great Bear Magmatic zone in the Wopmay orogeny (Cook, 2011).The age discrepancy may be attributed to the time difference between stages of a Wilson cycle; for example, formation of the protolith oceanic crust, eclogitization and incorporation into the cratonic mantle lithosphere during subduction (e.g.Shirey & Richardson, 2011).Furthermore, if we consider the isotopic composition of clinopyroxene from Jer 3 to represent that of the Paleoproterozoic protolith oceanic crust (and, by proxy, the mantle source), then we must account for in-growth of radiogenic Pb between oceanic crust formation and clinopyroxene growth during eclogitization.Considering extraction of oceanic crust from the depleted mantle at c. 2•2 Ga, possibly followed by seawater alteration in which hydrothermal alteration preferentially strips Pb over U, increasing both the U/Pb (and thus the m-value) of the altered crust (Chauvel et al., 1992), the oceanic crust would have c.200 Myr of radiogenic Pb in-growth prior to eclogitization.Fluid and/or partial melt extraction that accompanied subduction would preferentially remove U relative to Pb (Klemme et al., 2002), lowering both the bulk-rock eclogite and thus clinopyroxene U/Pb (50•15 and 50•07, respectively, for the Jericho eclogites), and thus significantly slow down isotopic in-growth, effectively freezing in the Pb isotopic ratios.
Assuming m values for 'normal' oceanic crust (m ¼8) and HIMU-type oceanic crust (m ¼22), the 206 Pb/ 204 Pb value increases between 0•25 and 0•7.This small change is La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu almost encompassed by the large error ellipses on the clinopyroxene Pb data in Fig. 5 (Schmidberger et al., 2005) have Hf model ages between 2•1 and 2•3 Ga and U^Pb ages of 1•9 Ga.In both these studies, the older model ages are interpreted to mark oceanic crust (protolith) formation whereas the younger ages mark subduction and eclogitization of the oceanic crust c. 200 Myr later.As demonstrated above, 200 Myr of isotopic in-growth does not substantially change the Pb isotope ratios and age interpretations.

Insights from Sr isotopes
Further insight into potential isotope mixing processes is gained from the Sr isotope systematics of the Jericho eclogites.The 87 Sr/ 86 Sr compositions of the eclogite clinopyroxene range between 0•7032 and 0•7052, but a large proportion of the data are c.0•7040.If protolith formation is assumed to have occurred at c. 2•2 Ga, then it is reasonable to assume that the eclogite clinopyroxene may preserve the Sr isotopic compositions of Paleoproterozoic oceanic lithosphere, given that the radiogenic in-growth of Sr isotopes during the 200 Myr between oceanic crust extraction from the upper mantle and eclogitization during subduction will change the 87 Sr/ 86 Sr values by only 0•0002 (using Rb/Sr ratios of unaltered and seawateraltered oceanic crust; Bach et al., 2001).Using the parameters given by Workman & Hart (2005), the 87 Sr/ 86 Sr ratio of oceanic crust extracted from depleted MORB mantle at c. 2 Ga is c. 0•7010, which is clearly much lower than the values observed in the Jericho eclogites.The oxygen isotope compositions of some Jericho eclogites indicate that seawater alteration may have affected some protoliths, and it is possible that this alteration could have caused the protoliths to assume the Sr isotopic composition of Paleoproterozoic seawater (0•7040 at c. 2 Ga; Veizer & Compston, 1976) in a similar fashion to the c. 2 Ga eclogites from the Diavik kimberlite (Schmidberger et al., 2007).However, there are still some Jericho eclogite clinopyroxenes with 87 Sr/ 86 Sr values significantly higher than 0•7040 that cannot reasonably be explained by this process.Mixing with a more radiogenic metasomatic component, comparable with the Jericho kimberlite, may explain some of the variation in both the Sr and Pb isotopic compositions of the Jericho eclogites.It is interesting that clinopyroxenes from the peaked eclogites largely have more radiogenic Sr isotope compositions compared with the depleted eclogites, and the same pattern is also be observed in the Pb isotope compositions (Fig. 5).This may suggest that the peaked eclogite group was affected to a greater extent by later metasomatic events.

The effect of kimberlite-related metasomatism
Figure 15 explores Sr^Pb isotope mixing relationships between (seawater-altered) Paleoproterozoic oceanic crust and potential metasomatic components.The mixing models require several components, including 'pristine' 2•2 Ga oceanic crust, seawater-altered c. 2 Ga oceanic crust, and a metasomatic agent, which we interpret to be the host Jericho kimberlite, or a similar 'pre-kimberlitic' fluid or melt (see Kopylova et al., 2009).All model values and sources are detailed in Table 7 and the captions to Figs 5 and 15.The Sr and Pb isotopic compositions of the Jericho clinopyroxenes define a diffuse, positively sloped array between relatively unradiogenic and more radiogenic Sr and Pb end-members (Fig. 15).Mixing between unaltered 2•2 Ga oceanic crust and the kimberlite-like metasomatic agents produces a mixing curve that does not correspond to the eclogite clinopyroxene because of the low 87 Sr/ 86 Sr of the oceanic crust.However, if a seawater-altered protolith is used (with more radiogenic initial Sr and a slightly more radiogenic Pb isotopic composition), mixing with the kimberlite produces mixing curves that approximate the array shown by the Jericho eclogites.From Fig. 15, regardless of whether the 206 Pb/ 204 Pb ratios for unaltered or altered oceanic crust are used, it appears that 51 to 5% mixing of the kimberlitic metasomatic agent (with the higher 87 Sr/ 86 Sr ratio) provides the best approximation of the eclogite data.In general, the depleted eclogites require less than 1% kimberlite mixing, whereas the peaked eclogites require more kimberlite metasomatism, up to 5%.Several of the diamond-bearing high-MgO eclogites (diamonds in Fig. 15) from Smart et al. (2009), however, appear to require much higher amounts of the metasomatic agent (420%) to produce the more radiogenic compositions.However, addition of 420% kimberlite would produce decidedly non-basaltic compositions, and as such, Smart et al. (2009) attributed the radiogenic isotopic composition of the high-MgO diamond-bearing eclogites to an early LREE-enriched metasomatic event, possibly associated with diamond formation.

Potential Mesoproterozoic metasomatic event
Age information derived from eclogitic diamond inclusions and eclogite xenoliths from Slave craton kimberlites shows peaks in the Paleoproterozoic, at c. 2 Ga (Schmidberger et al., 2005(Schmidberger et al., , 2007;;Aulbach et al., 2009a), whereas peridotitic diamond inclusions show peaks in the Neo-and Mesoarchean at c. 2•7^3 Ga (e.g.Irvine et al., 2003;Aulbach et al., 2009b) and, significantly, ages 43 Ga are rare (e.g.Westerlund et al., 2006;Aulbach et al., 2009b).In contrast, peridotite and zircon-bearing eclogite xenoliths from the Jericho kimberlite display a noteworthy abundance of Mesoproterozoic 1•3 Ga model ages (Irvine et al., 2003;Heaman et al., 2006), which have been attributed to modification of the northern Slave cratonic mantle by the 1•27 Ga Mackenzie Large Igneous Event (Le Cheminant & Heaman, 1989).The abundance of Mackenzie ages from Jericho mantle material compared with the effective absence from the central Slave mantle is probably due to the proximity of Jericho to the focal point of the Mackenzie intrusions in the northernmost Slave craton.The Mackenzie event involved large-scale mafic magmatism for which a mantle plume at the base of the lithosphere is thought to be the impetus (Le Cheminant & Heaman, 1989).The presence of a thermal anomaly at the base of the northern Slave cratonic mantle may have also caused volatiledriven incipient melting and the generation of lowvolume, silica-poor, volatile-rich melts, such as carbonatite-or lamproite/kimberlite-like melts (e.g.Dalton & Presnall, 1998;Wallace & Green, 1988;Ernst & Bell, 2010) that would be extremely reactive with the cratonic mantle (e.g.Tappe et al., 2012).The impact of such metasomatic events on cratonic mantle has been well studied in peridotite systems (e.g.Rudnick et al., 1993;Klemme et al., 1995), whereas less is known about the different effects in the eclogitic system (e.g.Jacob et al., 2009).The Jericho eclogites of this study contain secondary veins of calcite, apatite and phlogopite, which are probably the product of kimberlite infiltration.In contrast, both the high-MgO diamondbearing and zircon-bearing eclogites from Jericho contain macrocystal phlogopite that is attributed to carbonatitedriven metasomatism (Smart et al., 2009).Combined with Sm^Nd isotopic evidence from the zircon-bearing eclogites, this metasomatism has been hypothesized to be associated with the Mesoproterozoic Mackenzie event (Heaman et al., 2006).Thus, although there is no resolvable isotopic evidence that clearly indicates Mackenzie-age metasomatism in the Jericho eclogites of this study (i.e.no 1•3 Ga Pb model ages are observed), and kimberlite^eclogite mixing can account for the eclogite Sr^Pb isotope systematics, no relationships exist between Sr and Pb concentrations and isotopic ratios ( 87 Sr/ 86 Sr and Sr, r 2 ¼ 0•33; 206 Pb/ 204 Pb and Pb, r 2 ¼ 0•09).Although it may be over-optimistic to expect clear geochemical correlations in mantle-derived samples, the complete lack of any relationship could indicate that the eclogite system was disturbed by multiple metasomatic events during their storage in the Slave cratonic mantle.
The potential isotopic effect(s) of a 1•3 Ga Mackenziedriven metasomatic event are estimated in Fig. 15.The nature of this metasomatic event is difficult to constrain, and as such we have adapted three separate approaches.The first approach (Basaltic model) assumes that the Jericho cratonic mantle lithosphere was affected by the same mafic melts as now manifested in the Slave crust as widespread flood basalts, mafic dikes and the Muskox layered intrusion.The increasing Y, Zr, Ti and FeO contents with depth observed in Jericho peridotite xenoliths are consistent with melt (basaltic)-metasomatism (Griffin et al., 2004b).Constraints for this Basaltic model are taken from the Muskox layered intrusion from the northern Slave (Purves, 2007).The second (Hydrous model) and third (Carbonatite model) approaches are based on the assumption that low-volume, volatile-rich melts, which are more likely to be generated at plume peripheries (e.g.Wyllie, 1988), would be a better approximation of the thermal and melting regime present in the 1•3 Ga Jericho cratonic mantle lithosphere.Hydrous melting of peridotite produces lamproitic-type melts (e.g.Foley, 1993), whereas carbon-fluxed melting of peridotite produces carbonatitelike melts (e.g.Wallace & Green, 1988).Evidence for the influx of hydrous melts may be inferred from the presence of phlogopite and MARID-like geochemical signatures in Slave mantle xenoliths (Aulbach et al., 2007;Smart et al., 2009), whereas evidence of carbonatite-driven metasomatism is provided by carbonatite-like inclusions in kimberlite-derived megacrysts and diamonds from central Slave kimberlites (van Achterberg et al., 2002;Klein-BenDavid et al., 2007).For both the Hydrous and Carbonatite models, the metasomatic agents are estimated from Canadian Mesoproterozoic lamproites (Tappe et al., 2007) and carbonatites (Bell & Blenkinsop, 1989;Kwon et al., 1989).Details of the geochemical and isotopic composition of the models are found in Table 7 and the caption of Fig. 15.
Figure 15 shows the estimated effects of a Mackenzierelated metasomatic event.Mixing of the protolith with either basaltic melts (Basalt model) or low-volume, hydrous melts (Hydrous model) produces arrays with relatively unradiogenic 206 Pb/ 204 Pb values that do not correspond to the isotopic compositions of the Jericho eclogites.These metasomatic agents are further dismissed because the Jericho eclogites do not preserve any strong features of basaltic melt-metasomatism; specifically, the garnet has low TiO 2 contents (50•15 wt %) and lacks strong enrichments in Zr (520 ppm) (Fig. 9; e.g.Griffin et al., 2004a).By comparison, mixing of the protolith with the model carbonatite produces arrays with less radiogenic 87 Sr/ 86 Sr values than observed in the Jericho eclogites.Therefore, straightforward mixing of an oceanic crustal protolith in any of the Mesoproterozoic metasomatic models does not produce the observed Sr^Pb isotope systematics of the Jericho eclogites.Although the protoliths of the mantle xenoliths have long residence times in the cratonic mantle, and the probability is high that those protoliths have been affected by more than one metasomatic event (e.g.three-component mixing in Fig. 15), it appears that in the case of the Jericho eclogites the simplest explanation for the isotopic data involves metasomatism of the Paleoproterozoic oceanic protoliths by an agent similar to the host Jericho kimberlite (shaded arrow in Fig. 15).

Relationship to tectonomagmatic events in the Slave craton
Cratons such as the Kaapvaal craton in southern Africa and the North Atlantic craton in West Greenland^northeastern Canada preserve excellent links between Archean crustal growth, subduction events and the formation of eclogites (AE diamonds) in the cratonic mantle (e.g.Shirey et al., 2004;Wittig et al., 2010;Tappe et al., 2011).However, Neoarchean eclogitic mantle material, which is documented ubiquitously from the Kaapvaal craton, is thus far completely absent from the Slave craton in NW Canada.Instead, abundant Paleoproterozoic, c. 1•8^2•2 Ga eclogite xenoliths and eclogite-type diamond inclusions are known from kimberlites from both the central and northern part of the Slave craton (e.g. this study; Schmidberger et al., 2005Schmidberger et al., , 2007;;Aulbach et al., 2009aAulbach et al., , 2009b)).The apparent absence of older eclogitic material from the Slave craton is interesting in light of the evidence for a Neoarchean craton amalgamation event (Davis et al., 2003) that could have emplaced eclogitic material into the cratonic mantle and perhaps contributed to diamond formation at this time (e.g.Smart et al., 2011).Because glimpses of Neoarchean magmatic events are recorded in Slave peridotite xenoliths and P-type diamond inclusions (e.g.Irvine et al., 2003;Aulbach et al., 2009b), the apparent absence of a Neoarchean eclogitic record may be due to purging of the older, dense eclogites from the cratonic mantle root (e.g.Pearson & Wittig, 2008;Tappe et al., 2011) or to the heterogeneous sampling patterns of kimberlites (Aulbach et al., 2009a).Nevertheless, subduction associated with the Paleoproterozoic Wopmay orogeny appears to have been an active contributor of significant amounts of basaltic and volatile (e.g.carbon) components to the Slave cratonic mantle, which is manifested in the occurrence of c. 2 Ga eclogite xenoliths and diamond inclusions (this study; Schmidberger et al., 2005Schmidberger et al., , 2007;;Aulbach et al., 2009a;Smart et al., 2011).Furthermore, evidence for Mesoproterozoic metasomatism of the Slave cratonic root that is associated with the Mackenzie Large Igneous Event is recorded in the northern Slave mantle lithosphere, where c. 1•3 Ga model ages have been obtained for some Jericho peridotite xenoliths (Irvine et al., 2003) and eclogite xenoliths (Heaman et al., 2006).This Mesoproterozoic overprint, although not clearly recorded in the eclogites of this study, may nevertheless have contributed to cryptic Sr and Pb isotopic modifications.The integrated studies of Jericho mantle materials clearly suggest that major crustal events that affected the post-Archean Slave craton also significantly affected the Slave mantle lithosphere.Although the Slave cratonic mantle appears to have either lost or concealed a significant part of its Archean eclogite memory, it provides strong evidence for the operation of Paleoproterozoic subduction-driven modification processes that occurred during the assembly of the Laurentia supercraton (Hoffman, 1989).

Fig. 1 .
Fig. 1.Simplified geological map of the Slave craton of northwestern Canada, modified from Tappe et al. (2013).The Jericho and Muskox kimberlites are located in the northern Slave craton (indicated by stars), c. 150 km north of the Lac de Gras kimberlite field that hosts the Diavik and Ekati diamond mines.Based on xenocrystic garnet compositions, the Slave craton has been divided into Northern (N), Central (C) and Southern (S) mantle domains, which are delineated by NE^SW-trending dashed lines (from Gru« tter et al., 1999).

Fig. 2 .Fig. 3 .
Fig. 2. Ternary diagram showing the Ca^Mg^Fe compositions of garnet from Jericho and Muskox eclogite xenoliths.Group B diamond eclogites include two xenoliths from the Muskox kimberlite (15 km SW from the Jericho kimberlite).Shown for comparison are Jericho Group A diamond-bearing eclogites from Smart et al. (2009, 2011), Jericho zircon-bearing eclogites from Heaman et al. (2006), and previously published Jericho eclogites from Kopylova et al. (1999a) and De Stefano et al. (2009).The field of 'Slave eclogites' includes garnet from eclogite xenoliths from the Diavik kimberlite(Aulbach et al., 2007;Schmidberger et al., 2007) located in the central Slave craton and Voyageur kimberlite(Smart et al., in preparation)  located in the northern Slave craton.The geochemical group A^B^C compositional divisions are fromColeman et al. (1965).

Fig. 4 .
Fig.4.Chondrite-normalized REE patterns for garnet (a^c) and clinopyroxene (d^f) from Groups B, C and diamond-bearing Group B eclogites, as defined in Figs2 and 3. Grey fields in (c) and (f) represent the REE pattern for garnet and clinopyroxene from high-MgO Group A Jericho eclogites fromSmart et al. (2009).Normalization values are fromMcDonough & Sun (1995).

C
A L C U L AT E D W H O L E -RO C K C O M P O S I T I O N S illustrates the MgO vs SiO 2 , Al 2 O 3 , CaO, FeO and Na 2 O (wt %) and Ni (ppm) variations in the calculated eclogite bulk-rocks.A negatively sloping trend is observed between MgO and Al 2 O 3 in Fig. 6a, whereas MgO and Ni form a positively sloping trend in Fig. 6f.No strong correlations are observed between MgO and CaO, FeO, SiO 2 and Na 2 O (Fig. 6b^e).The Jericho eclogite bulk-rocks have broadly similar MgO, Al 2 O 3 , CaO, Na 2 O and Ni contents to modern altered oceanic gabbros (from Bach et al., 2001), but are markedly poorer in SiO 2 and enriched in FeO relative to the gabbros.At comparable MgO contents, the Jericho eclogites have similar CaO, FeO, Na 2 O and Ni contents to Archean basalts and picrites, but higher Al 2 O 3 and lower SiO 2 contents.

Fig. 7 .
Fig. 7. Calculated whole-rock REE compositions of the Jericho and Muskox eclogite xenoliths.(a) Representative patterns for the three new trace-element pattern types introduced for the Jericho and Muskox eclogites.(b) Peaked eclogites have negatively sloping, depleted LREE and mildly depleted to flat HREE patterns with distinctive positive Eu anomalies (Eu/Eu* ¼ 1•3^1• 7).(c) Depleted eclogites have overall higher contents of HREE, stronger depletions in the LREE and smaller Eu anomalies (Eu/Eu* ¼ 0•9^1•3) compared with the peaked eclogites.(d) Sloped eclogites have overall positive slopes in the REE, and are slightly depleted in LREE compared with the HREE.All Group B diamond-bearing eclogites fall into the sloped group.Symbols in (b)^(d) reflect the garnet composition classification from Fig. 2. Shaded field in (d) represents the composition of Jericho high-MgO diamond-bearing eclogites from Smart et al. (2009).Chondrite normalization values are from McDonough & Sun (1995).
U Nb Ta La Ce Pr Pb Sr Nd Sm Zr Hf Eu Ti Gd Tb Dy Y Ho Er TmYb Lu

Fig. 9 .
Fig. 9. Evaluation of the effects of metasomatism on the Jericho and Muskox eclogites based on garnet Zr (ppm) and clinopyroxene Ce (ppm) contents.Compared with garnet and clinopyroxene from modally metasomatized eclogite suites that have associated evidence of cryptic metasomatism (e.g.phlogopite eclogites from Kimberley, Jacob et al., 2009; phlogopite þ diamond eclogites from Jericho, Smart et al., 2009), the Jericho and Muskox eclogites of this study have very low contents of incompatible elements.

Fig. 13 .
Fig.13.Trace-element modeling of the petrogenesis of the Jericho peaked and depleted eclogites.(a) Comparison of peaked eclogites with modeled gabbro.The model gabbro composition was calculated to have formed from a 6% fractional partial melt of depleted MORB mantle (DMM) [starting composition and fraction of melting fromWorkman & Hart (2005)] using the partition coefficients ofKelemen et al. (1993).The gabbro composition was calculated assuming 90% fractional crystallization of this melt using the partition coefficients for plagioclase, pyroxenes and olivine fromNorman et al. (2005).The modal abundance of plagioclase, clinopyroxene, orthopyroxene and olivine (53%, 20%, 0%, 27%) used for the calculated gabbro was determined from the average CIPW norms calculated for the whole-rock compositions of peaked Jericho eclogites.Field of oceanic gabbros is fromBach et al. (2001) and representative ophiolite olivine gabbro is fromBenoit et al. (1996).(b) Jericho depleted eclogites are modeled as residual peaked eclogites after partial melt extraction at high pressure.Incremental 5, 10 and 20% batch partial melts were extracted from the average peaked Jericho eclogite using partition coefficients for gabbroic-protolith eclogite/melt fromYaxley & Sobolev (2007) and a mode of 50% garnet and 50% clinopyroxene.Chondrite normalization values are fromMcDonough & Sun (1995).

Table 1 :
Jericho and Muskox eclogite petrography . The REE patterns for Group B, C

Table 2 :
Major-element composition of garnet, clinopyroxene and calculated whole-rocks of Jericho and Muskox eclogites

Table 3 :
Trace-element composition of garnet and clinopyroxene fromJericho and Muskox eclogites

Table 5 .
The d 18 O values determined for garnet from both Group B and C eclogites vary between 5•2 and 5•8ø, excluding one kyanite-bearing Group C eclogite (10-13) with higher d 18 O values of 6•36 The oxygen isotope composition of garnet from Jericho eclogite Mx8A [a zircon-bearing eclogite from Schmidberger et al. (2005) and Heaman et al. (2006) with a U^Pb age of 1•9 Ga] has a d 18 O value of 5•6ø.

Table 5 :
Oxygen isotope compositions of garnet and clinopyroxene fromJericho eclogites

Table 6 :
Calculated whole-rock trace-element composition of Jericho and Muskox eclogites N /(Gd N þ Sm N ); Sr* ¼ 2Sr PM /(Pr PM þ Nd PM).Subscripts N and PM indicate normalization to chondrite and primitive mantle respectively from McDonough & Sun (1995).Class., classification of eclogites according to wholerock chondrite-normalized REE pattern (see Figs 7 and 8 and text).
demonstrated that addition of a hypothetical eclogite melt product (e.g.tonalite) can (Schmidberger et al., 2007)the overall interpretation that the Jericho eclogites have Paleoproterozoic formation ages.Applying the common Pb model age method to the Jericho eclogites requires that the smearing of the remaining Jericho Pb data towards more radiogenic isotope compositions may be due to mixing of Pb between Paleoproterozoic mantle and Phanerozoic kimberlite sources, and therefore the c. 1•7 Ga age calculated from the Pb secondary isochron may be meaningless.Regardless of whether the 1•7 AE 0•3 Ga secondary isochron or the 2•2 Ga minimum model age is preferred for the Jericho eclogites, the overall interpretation that the Jericho eclogites formed in the Paleoproterozoic remains unchanged.Two other eclogite suites from the Slave craton provide geochronological evidence that may capture different stages of Wilson cycle processes.Eclogites from the Diavik kimberlite(Schmidberger et al., 2007)have older Pb model ages (2•4^2•3 and 2•1Ga) but a younger Lu^Hf errorchron age (2•1 AE0•3 Ga), whereas zircon-bearing eclogites also from Jericho