Mapping the Structure and Metasomatic Enrichment of the Lithospheric Mantle Beneath the Kimberley Craton, Western Australia

The lithology, geochemistry, and architecture of the continental lithospheric mantle (CLM) underlying the Kimberley Craton of north‐western Australia has been constrained using pressure‐temperature estimates and mineral compositions for >5,000 newly analyzed and published garnet and chrome (Cr) diopside mantle xenocrysts from 25 kimberlites and lamproites of Mesoproterozoic to Miocene age. Single‐grain Cr diopside paleogeotherms define lithospheric thicknesses of 200–250 km and fall along conductive geotherms corresponding to a surface heat flow of 37–40 mW/m2. Similar geotherms derived from Miocene and Mesoproterozoic intrusions indicate that the lithospheric architecture and thermal state of the CLM has remained stable since at least 1,000 Ma. The chemistry of xenocrysts defines a layered lithosphere with lithological and geochemical domains in the shallow (<100 km) and deep (>150 km) CLM, separated by a diopside‐depleted and seismically slow mid‐lithosphere discontinuity (100–150 km). The shallow CLM is comprised of Cr diopsides derived from depleted garnet‐poor and spinel‐bearing lherzolite that has been weakly metasomatized. This layer may represent an early (Meso to Neoarchean?) nucleus of the craton. The deep CLM is comprised of high Cr2O3 garnet lherzolite with lesser harzburgite, and eclogite. The peridotite components are inferred to have formed as residues of polybaric partial mantle melting in the Archean, whereas eclogite likely represents former oceanic crust accreted during Paleoproterozoic subduction. This deep CLM was metasomatized by H2O‐rich melts derived from subducted sediments and high‐temperature FeO‐TiO2 melts from the asthenosphere.

• The Kimberley Craton has retained a thick (>220 km) thermally stable lithospheric root since the Mesoproterozoic • The lithospheric mantle comprises a depleted shallow layer (<100 km) and metasomatized deep layer (>150 km) separated by a diopside-depleted mid-lithosphere discontinuity • Diamondiferous lamproites at the craton margin are associated with eclogite and high levels of K 2 O metasomatism of the lithospheric mantle

East Kimberley Province
The East Kimberley Province contains >34 lamproite, kimberlite and ultramafic lamprophyre dykes and pipes (Figure 1).Most occur as small dykes that intrude the Paleoproterozoic Halls Creek Orogen, but several intrude Kimberley Basin sediments.Most dykes and pipes are inferred to be of Neoproterozoic age, but only Argyle, Bow Hill, and Mad Gap Yards have isotopic ages.The Argyle lamproite, and by association, the Lissadell Road and Seagull's Dyke lamproites, are the oldest with Argyle having an age of 1,177 ± 47 Ma (Pidgeon et al., 1989).

Central Kimberley Province
Six kimberlite pipes have been identified in the Aries kimberlite field which intrudes Kimberley Basin sediments in the southern-central part of the Kimberley Craton (Figure 1).The largest of these is the Aries pipe which comprises three lobes with a combined area of ∼19 ha.The other pipes (Athena, Helena, Persephone, and Niobe) are typically 1-2 ha.The Aries pipe has an Ar-Ar age of 815.4 ± 4.3 Ma (Downes et al., 2006).The Aries pipes are infilled by lithic-rich kimberlite breccias and intruded by macrocrystic phlogopite kimberlite, and minor olivine-phlogopite-richterite kimberlite, and late-stage macrocrystic serpentine-diopside ultramafic dykes (Downes et al., 2006;Edwards et al., 1992).Chromian spinel and Cr pyrope have been recovered from the pipes.Ramsay et al. (1994) identified two main groups of spinels: a Ti-poor (<1 wt % TiO 2 ) and Cr-rich (>55 Cr 2 O 3 wt %) group which is typically cross-cut and rimmed by later more Fe and Ti-rich spinels thought to have formed by metasomatic/magmatic events in the mantle, perhaps related to kimberlite formation.

Note.
See Supporting Information S1 and S2 for raw data.inf is inferred.Table 1 Continued

West Kimberley Province
The West Kimberley Province lies at the southwest margin of the Kimberley Craton extending from the Proterozoic Wunaamin Miliwundi Orogen to the Fitzroy Trough (Figure 1).The province comprises more than 150 Miocene-age intrusions clustered in three main fields with a few isolated smaller volcanic centers in the Fitzroy Trough and the Lennard Shelf to the north (Jaques, 1986(Jaques, , 2006;;Jaques et al., 1984).The Ellendale field in the north lies within the Lennard Shelf and comprises 55 lamproites that occur as poorly exposed volcanic pipes of olivine lamproite and associated volcanic vents and stocks of leucite lamproite (Jaques, 1986).The Calwynyardah Field lies 40 km further south and comprises buried volcanic pipes filled with pyroclastic olivine lamproite and olivine-leucite lamproite.The Noonkanbah field lies 30 km south of the Calwynyardah Field in the center of the Fitzroy Trough and consists of leucite lamproite plugs and pipes, with minor olivine-leucite lamproite.
The lamproites are Miocene in age, with K-Ar and more recent Ar-Ar analyses indicating ages ranging from ∼21-23 Ma in the north to ∼19-20 Ma in the Noonkanbah field in the south, except for the large Walgidee Hills lamproite dated at 17.4 Ma (Jaques et al., 1984;Phillips et al., 2022).The olivine lamproites contain variable proportions of olivine xenocrysts and olivine-rich micro-xenoliths derived from mantle peridotite (Jaques, 1986;Jaques & Foley, 2018) and xenoliths of spinel lherzolite are present in some pipes (notably Ellendale 7).Ellendale 4, 7, 9, 17, 18, and 33 are concealed volcanic vents of variable sizes and shapes ranging in size from ∼6.8 ha (Ellendale 33) to the large multiphase complex pipes of Ellendale 4 (76 ha) and 9 (47 ha).All comprise basal layers of tuffs and lapilli tuffs, commonly with abundant accidental quartz grains derived from the underlying country rock sandstones, overlain by, and in some pipes interfingering with, vent-filling massive olivine lamproite and phlogopite-olivine lamproite.Ellendale 2 comprises several thin (≤5 m) sills of phlogopite-olivine lamproite.The Water Reserve pipe is a small (3.2 ha) vent comprised of quartz-rich tuffs and lapilli tuffs overlain by olivine-leucite lamproite.The Calwynyardah pipe at 124 ha is, after Walgidee Hills, the largest known pipe in the West Kimberley Province.It is composed entirely of tuffs, mostly pyroclastics phlogopite-olivine lamproite lapilli tuff.Layman's Bore East is an irregular shaped pipe of 103 ha comprised of a thick sequence of olivine lamproite tuffs and volcaniclastic tuffs.Laymans Bore West comprises two small vents (12 and 16 ha) of pyroclastic dominated by olivine-leucite lamproite tuff and lapilli tuff.The Walgidee Hills lamproite is the largest (∼2.4 km diameter) and youngest lamproite in the West Kimberley Province.It is zoned in terms of grain size, mineralogy, and mineral and rock composition from porphyritic olivine lamproite at the margin of the intrusion through medium-grained lamproite composed of olivine, Ti-phlogopite, diopside, leucite and K-Ti-richterite with accessory priderite, perovskite, apatite, and wadeite to coarse gained and pegmatitic lamproite rich in priderite, jeppeite, perovskite, apatite, wadeite and noonkanbahite at the center of the body.Mantle fragments recovered from mineral concentrate are dominated by Cr spinel with subordinate Cr diopside and Cr pyrope and traces of diamond (Jaques, 1986).

Analytical Methods
A data set of approximately 5,000 garnet and Cr diopside analyses has been used to study the Kimberley Craton CLM.With the exception of Argyle xenolith data taken from Jaques et al. (1990Jaques et al. ( , 2018) ) and Luguet et al. (2009), all Cr diopsides and garnets were sourced from heavy mineral concentrate and/or diamond inclusions.Our data set comprises more than 1200 Cr diopsides and more than 1300 pyrope and eclogitic garnets analyzed in this study.Previously unpublished analyses included garnet xenocrysts from Argyle, Ellendale, Bow Hill, Skerring, and Aries (Ramsay, 1992; Table 1).We include published analyses of garnet from Wyatt et al. (1998) and W. L. Griffin et al. (1999), which included samples from Argyle, Bow Hill, Ellendale, Pteropus, Seppelt and Skerring and the Mad Gap Yards lamprophyre from Downes et al. (2023) (Table 1).Garnet inclusions in diamonds from Argyle and Ellendale were taken from Jaques et al. (1989bJaques et al. ( , 1984)), W. L. Griffin et al. (1988), Sobolev (1989), andStachel et al. (2018).Garnet and Cr diopside xenocrysts measured as part of this study were typically homogenous and lacked metasomatic zonation between the core and rim.A small portion of our data set (∼2.5%) was comprised of rare metasomatically zoned Cr diopsides.These samples contained geochemically enriched rims that were likely caused by melt-rock interactions with the host kimberlite/lamproite melt.For the purpose of this study, the metasomatically altered regions of the xenocrysts were avoided during microanalysis.
Quantitative analyses of major and minor oxides on the newly acquired data set were performed using calibrated wavelength dispersive spectroscopy (WDS) on a JEOL JXA-8530F+ electron probe microanalyzer (EPMA) at the Centre for Advanced Microscopy, the Australian National University.A 15 or 20 kV accelerating voltage, 20 nA beam current and 5 μm beam diameter were used for most mineral analyses.Counting times were 20 s (peak and background) for major elements and 30-40 s for minor elements.The calibrations were made using synthetic and natural ASTIMEX Mineralogy standards: Na (albite), Mg (periclase and olivine), Al (sanidine and pyrope), Si (sanidine and diopside), Ca (diopside), Ti (rutile), Cr (chromite), Mn (rhodonite), Fe (hematite and magnetite), K (sanidine), and Ni (pentlandite).Matrix correction was performed using the XPP model (Pouchou and Pichoir, 1991).The limits of detection were typically 70-180 ppm, equivalent to 0.01-0.03wt % oxide.
The concentrations of 40 trace element isotopes (see Supporting Information S1 and S2 for list) in garnet and Cr diopside were measured by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) using an Agilent Technologies 7700x quadrupole mass spectrometer coupled to a Coherent Scientific Compex Pro 110 (λ = 193 nm) laser ablation system at the Research School of Earth Sciences, Australian National University.Analyses were done using a HelEx dual-volume sample cell with the laser operating at a 5 Hz pulse rate, 80 mJ output energy, and 47 μm diameter circular spot, using 0.5 L/min He and 0.9-1.0L/min Ar sample gas to transport ablated material to the ICP-MS.Data acquisition involved a 40 or 45-s ablation time and a background (Ar-gas blank) pre-and post-ablation time of 20 or 25 s.Garnet and Cr diopside silicon values determined by EPMA were used as the internal standard and NIST reference glass SRM 610 (NIST 610) was used as the primary standard (Jochum et al., 2011).Reference glasses BCR-2G and NIST 612 were monitored as secondary standards (Jochum et al., 2011).Standard measurements bracketed every 10 or 20 analyses.The data obtained were reduced using the Iolite v2.3 software package (Paton et al., 2011).Time resolved ICP-MS trace element data for representative standard measurements taken from several sessions (BCR-2G, NIST 610, and NIST 612) are reported in Supporting Information S1 and S2.All raw data from this study and previous published analyses are reported in Supporting Information S1 and S2.

Geothermobarometry and Paleogeotherm Modeling
Equilibration pressures (P) and temperatures (T) of Cr diopsides were determined using the experimentally calibrated single-grain Cr-in-clinopyroxene geobarometer of Sudholz, Yaxley, Jaques, and Brey (2021) and the enstatite-in-clinopyroxene geothermometer of Nimis and Taylor (2000), with data quality and filtering protocols following Ziberna et al. (2016).The filtering protocols were used to removed Cr diopsides that did not equilibrate with garnet in a pyroxenitic or peridotitic system.The authors acknowledge that, despite the use of these filters, it is possible that some garnet-spinel and/or spinel bearing-peridotitic Cr diopsides have been inadvertently included in our data set (see Supporting Information S1 and S2).To address this issue, we have removed some additional Cr diopside analyses with obvious Al 2 O 3 enrichment and flat rare earth element (REE) patterns with HREE enrichment diagnostic of equilibration in a garnet-free spinel bearing source.Future work is needed to better understand the applicability of Cr diopside filtering methods (i.e., Sudholz, Yaxley, Jaques, & Brey, 2021;Ziberna et al., 2016) to chromite-bearing systems or deep-seated garnet-free lithotypes from the cratonic lithosphere.All PT estimates in this study were calculated using PTEXL (https://cms.eas.ualberta.ca/team-diamond/downloads/) (maintained by T. Stachel).The calibration set for single grain clinopyroxene geothermobarometry has an inferred uncertainty of approximately ±35-50℃ and ±0.3-0.45GPa based on error propagation and application to natural data sets, with the highest uncertainties associated with pressure estimates above 6 GPa.The Sudholz, Yaxley, Jaques, and Brey (2021) calibration of the Cr-in-clinopyroxene geobarometer is favored over the earlier calibration by Nimis and Taylor (2000) because of its improved precision and accuracy above 4.5 GPa.It is also preferred over the recent empirical correction proposed by Nimis et al. (2020), who used P estimates from the Carswell (1991) Al-in-orthopyroxene geobarometer (modified after Nickel and Green (1985)) for natural xenoliths to correct for P-underestimation above 4.5 GPa.Although this correction yields improved results on high-P (>4 GPa) xenoliths relative to the original Nimis and Taylor (2000) calibration (see Figure 6, Nimis, 2022), the uncertainties on P estimates, which undoubtably propagate from the empirical fit, remain poorly understood, but are estimated to be in excess of 0.75 kbar based on the propagation of uncertainties of 0.3 GPa for each of the geobarometers (Carswell, 1991;Nimis & Taylor, 2000).Estimates of equilibration T for pyrope garnet were made using the experimentally calibrated single-grain Ni-in-garnet geothermometer of Sudholz, Yaxley, Jaques, and Chen (2021), using garnet Ni concentrations measured by LA-ICP-MS and a recommended olivine Ni concentration of 3,000 ppm.This calibration is preferred over the empirical calibration of Ryan et al. (1996) because of its improved precision and accuracy across more of the PT range of garnet in the CLM (see Sudholz, Yaxley, Jaques, and Chen ( 2021)).As shown in Sudholz, Yaxley, Jaques, and Chen (2021), the fixed olivine Ni concentration has a negligible effect on the calculated T between 2,500 and 3,500 ppm.Published garnet analyses taken from W. L. Griffin et al. (1999) include a small number of garnets from Argyle and Bow Hill with acceptable cation totals but analytical totals outside the normal acceptable range (98-102 wt%): these have been included for classification purposes only.The vast majority of analyses, including most of the newly acquired data set, have oxide totals between 98.5 and 101.5 wt% (Supporting Information S1 and S2).The cut-off of 98-102 wt% allows us to include several legacy data sets for important Proterozoic intrusions (i.e., Skerring and Argyle) from exploration programs (Ramsay pers.Comms).Most of the garnet analyses with oxide totals in the range of 98-102 wt% were not used for geothermobarometry (Supporting Information S1 and S2).The equilibration P of garnets were determined by referencing Ni-in-garnet T estimates to the local paleogeotherm defined by the Cr diopside xenocrysts.
The Mg# of olivine (Mg# olv = Mg/(Mg + Fe)) coexisting with garnet in peridotite was determined by inverting the garnet-olivine Fe-Mg exchange geothermometer of O' Neill and Wood (1979) to solve for Ni-in-garnet T (see Sudholz et al. (2022b)).
Paleogeotherms for each intrusion were modeled in FITPLOT (Mather et al., 2011) using the PT estimates from Cr diopsides.Crustal thicknesses at each intrusion were taken from the AuSREM Moho model (Aitken et al., 2013;Salmon et al., 2013) (http://rses.anu.edu.au/seismology/AuSREM/AusMoho/).For most pipes, crustal thicknesses varied between 35 and 45 km.Paleogeotherms were calculated using a mantle isentrope of 1330℃, a value based on the thermal structure of cooling oceanic lithosphere and seismological evidence (Hoggard et al., 2020;Katsura et al., 2010;Levin et al., 2023;Mather et al., 2011;Richards et al., 2018).A 1330℃ value is also supported by the perturbed paleogeotherm defined by the Skerring xenocryst suite which is aligned along a 1300-1330℃ isentrope (see below).As will be discussed in greater detail below, the similarity in lithospheric and thermal architecture for Miocene and Mesoproterozoic intrusions indicates that the nature of the lithosphere across the craton has changed minimally over the past billion years.An estimate of the LAB for each intrusion was taken from the intersection point between the isentrope and paleogeotherm.The uncertainty for each paleogeotherm is defined by an error envelope calculated from the misfit of the iterated calculated paleogeotherm to the input PT array, using a root mean square distribution of ΔT from the calculated paleogeotherm (Mather et al., 2011).All PT estimates for Cr diopsides have been projected onto the model conductive paleogeotherms of Hasterok and Chapman (2011) (see below).

Lithospheric Thickness and Mantle Heat Flow
Paleogeotherms and LAB depth estimates are similar for Miocene (Ellendale Field) and Mesoproterozoic (Argyle) lamproites (Figure 2).The depth to the LAB beneath intrusions within the West Kimberley Province ranges between 220 and 250 km (Figure 2), except for Ellendale 9 and Layman's Bore West.In these cases, the position of the paleogeotherm is biased by the overabundance of samples from the shallower sections of the CLM.There is little discernible difference in the LAB thickness estimates between those from the Ellendale Field in the north and for the Walgidee Hills lamproite in the northern Noonkanbah Field ∼100 km south.The thick lithosphere within the province is consistent with the presence of diamond and the seismic tomography models of  2023) and Magrini et al. (2023).The LAB thickness estimates for the East Kimberley Craton are constrained by PT estimates from the Mad Gap Yard (∼205 km) and Argyle (∼225 km) intrusions (Figure 2).The LAB estimate for Argyle is slightly deeper than previous estimates by Jaques et al. (2018), which used the Nimis and Taylor (2000) calibration of the Cr-in-clinopyroxene geobarometer which commonly underestimates pressures above 4.5 GPa (see Sudholz, Yaxley, Jaques, and Brey (2021) for discussion).The LAB estimate for the North Kimberley Province (∼245 km) is constrained by PT estimates for the Ashmore kimberlite (Figure 2).The Cr diopsides from the Skerring pipe define a perturbed paleogeotherm with a vertical PT array.
Single-grain PT estimates on Cr diopsides from the Kimberley Craton fall along conductive model geotherms corresponding to a surface heat flow of 37-40 mW/m 2 (Hasterok & Chapman, 2011;Supporting Information S1 and S2).The projected surface heat flows for Cr diopsides from the deeper Kimberley Craton CLM (>150 km) are similar for most kimberlite and lamproite provinces across the craton and suggest that the CLM has been thermally stable since the Mesoproterozoic (Supporting Information S1 and S2).The only exception are PT estimates from the Skerring kimberlite (Northern Kimberley Province), which define a perturbed geotherm that sits along the mantle isentrope (Supporting Information S1 and S2).Several small populations of Cr diopsides from the West Kimberley Province (notably Ellendale 4) have PT estimates within the shallow CLM (<100 km) that project onto slightly higher conductive geotherms, that correspond to surface heat flows closer to 40-45 mW/m 2 (Supporting Information S1 and S2).The slightly higher values may be caused by lithological variations within the shallow CLM that have adversely affected PT estimates on shallow CLM Cr diopsides.Alternatively, they may relate to slightly higher concentrations of radiogenic elements or thermal perturbations from magmatism.It is also possible that some or the more Al-rich Cr diopsides may not have equilibrated with garnet, although this is inconsistent with the results from data quality filtering.

Mantle Lithology Based on Garnet Composition
The concentration of Cr 2 O 3 and CaO in garnets from the Kimberley Craton defines a lherzolitic (G9) mantle with minor harzburgite (G10) and a smaller population of eclogite (G3) and pyroxenite (G4 and G5), consistent with previous studies (Jaques et al., 2018;Lucas et al., 1992) (Figure 3).Most G9 garnets fall within the low-Cr 2 O 3 (L-G9) and moderate-Cr 2 O 3 (M-G9) sub-fields, with rarer high-Cr 2 O 3 (H-G9) and G5 varieties occurring within the deep CLM (>150 km), particularly beneath the North Kimberley Province (Figure 3).Eclogitic garnets occur in the North (Skerring), West and East Kimberley Provinces but are most abundant in the East Kimberley Province (Figure 3).Garnets of this composition are the predominant inclusion type found in Argyle diamonds and are  et al. (2022b) and Grütter et al. (2004).H-G9 field occurs above 7 wt% Cr 2 O 3 .M-G9 field occurs between 7 and 3 wt% Cr 2 O 3 .L-G9 field occurs between 1 and 2.5 wt% Cr 2 O 3 .For descriptions of the garnet subclasses see Section 4.2 and Grütter et al. (2004).also present in Ellendale diamonds (Jaques et al., 1989b;Stachel et al., 2018).Pyroxenitic garnets are uncommon across the craton but occur in the West and East Kimberley Provinces and at the Skerring kimberlite (Figure 3).G1 garnets are abundant in the Skerring and Maude Creek kimberlites with much smaller populations present in the other pipes (Figure 3).Ni-in-garnet T estimates on pyrope garnet xenocrysts projected onto the local paleogeotherm show that the garnets were derived from 100 to 250 km depth with most derived from 130 to 230 km (Figure 4).The apparent absence of garnets from the shallow CLM is discussed in Section 6.2.
Garnets from the East Kimberley province have predominantly L-G9 and M-G9 compositions, but also comprise eclogitic (G3) varieties (Figure 5).The garnets from Argyle include a high abundance of crustal almandine (Lucas et al., 1992) but mantle garnets include L-G9 and M-G9 lherzolitic types with rare H-G9 type garnets with up to 11 wt % Cr 2 O 3 , eclogite and pyroxenite garnets (G3 and G4), minor low Cr megacrysts (G1) and titanian pyrope (G11 with up to 0.87 TiO 2 wt%), and rare low Cr sub-calcic G10 garnets (Figure 5).This contrasts with the overwhelming dominance of eclogitic inclusions in Argyle diamonds (Jaques et al., 1989b;Stachel et al., 2018; see below).The eclogitic garnets in heavy mineral concentrate lack the high Ti and Na contents of the high temperature eclogitic inclusions common in Argyle diamonds (Jaques et al., 1989b;Stachel et al., 2018) indicating that they were likely derived from eclogite domains higher in the lithosphere than the main eclogitic diamond source.The high temperature eclogitic inclusions at Argyle most likely reflect derivation from the bottom 50 km of the CLM (Stachel et al., 2018;Timmerman et al., 2019).Rare H-G9 and G10 high-Cr, low-Ca garnets occur as inclusions in Argyle peridotitic diamonds (Jaques et al., 1989b;Stachel et al., 2018) indicating derivation from Cr-rich lherzolite and harzburgite sources in the diamond stability field.In contrast, the garnet population from the nearby Bow Hill dykes, using a larger data set, indicates a high proportion of eclogitic garnets, more consistent with the eclogite-rich source of most Argyle diamonds (Figure 5).Again, the Bow Hill G3 garnets, with rare exception, have much lower Ti and Na than the eclogitic inclusions in Argyle diamonds indicating derivation from cooler and shallower mantle domains.The garnet population at the Maude Creek kimberlite has a high proportion of low Cr pyrope (G1), a significant population of G10 and medium and high Cr pyrope, but no eclogitic or pyroxenitic garnets (Lucas et al., 1992; Figure 5).These large differences in the relative proportions of eclogitic, peridotite and pyroxenite garnet sources suggest spatial variations in the CLM in this region occur on a scale of tens of kilometers.The data set also shows that, while there appear to be differences in the relative abundance of different G9 peridotitic garnets between locations, a common feature is the restriction of Ti-pyropes (≥0.8 wt % TiO 2 ) to lithospheric depths ≥190 km.
Garnets from the North Kimberley Province are of peridotitic Cr-pyrope composition (G9, G10) with a smaller population of low-Cr pyrope of the megacryst suite and a minor proportion of eclogitic and pyroxenitic garnets at the Skerring kimberlite pipe (Figure 5).The garnet populations of Ashmore, Pteropus, Seppelt and Bulgurri are dominated by G9 garnets, but Bulgurri also includes a significant population of G10 harzburgitic garnets (Figure 5) and a small number of high Mg# (∼88) Cr pyrope with low CaO (<2 wt %) suggesting they belong to an ultra-depleted harzburgite-dunite lithofacies.In contrast, the nearby Skerring pipe, in addition to the predominant low-Cr pyrope of the megacryst suite and lherzolitic G9 and less common G10 pyrope garnets, contains a smaller population of eclogitic and pyroxenitic garnets (Figure 5).High-TiO 2 (0.8-1.3 wt%) signatures in pyrope garnet are common within the deep CLM (≥150 km), particularly in the Skerring pipe.The relative abundance of Ti-rich pyropes and the high equilibration T of the sub-calcic Cr diopsides are notable features of Skerring pipe mantle xenocrysts (Figure 5).
Garnets from the West Kimberley Province are from the Ellendale Field (Figure 4).Pyrope garnets are mostly L-G9 and M-G9 types and are recorded throughout the mantle column between <120 and >200 km (Figure 5).Other features are the significant population of G10 low-Ca pyropes indicating a harzburgite source in the Ellendale CLM.Also, present in these pipes are pyropes with higher TiO 2 contents (≥0.5 wt %): these belong mostly to the low-Cr megacryst suite with a smaller proportion from the G11 class derived from near the base of the CLM (≥200 km depth).Both eclogitic and pyroxenitic garnet are abundant (Figure 5).The proportions of eclogitic garnets are greater in comparison to the North Kimberley Province, but less in comparison to the East Kimberley Province (Figure 5).The Ellendale eclogitic garnets have low TiO 2 (≤0.5 wt %) and Na 2 O (≤0.20 wt %) and are like those found as inclusions in the Ellendale eclogitic diamonds (Jaques et al., 1989b).Garnet-clinopyroxene Fe-Mg exchange geothermometry (Sudholz et al., 2022a) indicates that the Ellendale eclogitic inclusions equilibrated at lower T (1120-1220℃).The occurrence of high-TiO 2 peridotite and megacryst (G1 and G11 types) varieties along the base of the lithosphere is consistent with the North and East Kimberley populations of garnet (Figure 5).

Mantle Geochemistry-Chrome Diopside
Chrome diopsides were derived from garnet (±spinel) peridotite sources within an on-craton and/or cratonic margin setting.Some Cr diopsides with high Mg# and low Al 2 O 3 (2-4 wt %) may have been sourced from spinel  Chrome diopsides from the West Kimberley Province were sourced from ∼60 to 235 km depth (Figure 4).With the exception of Layman's Bore West, similar sampling intervals (as defined by Cr diopside PT estimates) were recorded across the entire West Kimberley Province (Figure 4).The shallow CLM is characterized by high Mg# (0.93-0.96) and low TiO 2 contents (mostly <0.3 wt %), and lower concentrations of Cr 2 O 3 and FeO in comparison with the deep CLM (Figure 6).Chrome diopsides from the shallow CLM have higher Al 2 O 3 and CaO, and lower K 2 O contents.Some Cr diopsides have elevated concentrations of Sc, Zn, and Y, and lower concentrations for Sr, Ga, Co, and Mn (Figure 7; Supporting Information S1 and S2).Primitive mantle (PM) normalized REE patterns for Cr diopsides range from highly LREE-depleted (La/Yb N < 0.5) to flat or slightly LREE-enriched (Figure 8; Supporting Information S1 and S2).In contrast, Cr diopsides from the deep CLM record enrichments in K 2 O (up to 0.4 wt %) and FeO (Figure 6).Deep CLM Cr diopsides record elevated concentrations of Zr, Sr and Ga, and lower concentrations of Sc, Zn and Y (Figure 7), as well as lower ratios of Ti/Eu and Zr/Y, and higher ratios of Ti/Al, V/Sc, La/Yb, and Sm/Er.Chrome diopsides from the deeper CLM show greater enrichment in LREEs compared to the shallow CLM (Figure 8).The highest LREE enrichments were recorded for samples that equilibrated near the base of the LAB (Figure 8).Chrome diopsides from the East Kimberley Province equilibrated between 60 and 210 km, although most grains equilibrated at 150-200 km (Figure 4).Shallow CLM samples occur as an isolated population at 60-80 km depth and within the mid-lithosphere (∼120 km) (Figure 4).Low-P Cr diopsides have higher concentrations of Al 2 O 3 and CaO and lower concentrations of FeO and MgO compared to high-P varieties (Figure 6).The Mg# of shallow Cr diopsides range between 0.93 and 0.94 and Cr 2 O 3 contents between 0.3 and 1.9 wt % (Figure 6).Chrome diopsides that equilibrated within the deep CLM (>150 km) have higher concentrations of FeO, MgO, and TiO 2 and lower Cr 2 O 3 (Figure 6).Chrome diopsides that equilibrated near the LAB recorded the highest concentrations of TiO 2 (Figure 6).Many Cr diopsides from Argyle are enriched in K 2 O (>0.2, up to 1.3 wt%) (Figure 6, Jaques et al., 1990Jaques et al., , 2018;;Luguet et al., 2009).Chrome diopsides in Argyle xenoliths derived from the deep CLM are strongly depleted in HREE and high-field strength elements (HFSE) but mostly moderately enriched in LREE and other large ion lithophile elements (LIL) elements relative to the PM (Jaques et al., 2018).
Chrome diopsides from the North Kimberley Province come from the Skerring and Ashmore pipes.The sampling interval of Cr diopsides occurred between ∼70 and 245 km and 875 and 1420°C, although most samples equilibrated between 150 and 200 km (Figure 4).As noted, a small sub-population of high-P Cr diopsides from Ashmore 1 and the low Cr (≤1 wt % Cr 2 O 3 ) sub-calcic diopsides from the Skerring pipe record anomalously high temperatures that plot along the mantle adiabat (Figure 2b).The Cr diopsides from the Ashmore pipes have Mg# of 0.91-0.94and define trends of decreasing Cr 2 O 3 and increasing FeO and (slightly) increasing TiO 2 with depth (Figure 6).Chrome diopsides from Ashmore have Mg# of 0.91-0.94,with the highest values also occurring at the base of the shallow CLM and the top of the deep CLM (Figure 6).In contrast, the sub-calcic diopsides from the Skerring pipe have uniformly low Mg# (0.87-0.89) and higher TiO 2 (0.14-0.44 wt %) (Figure 6).

Mantle Geochemistry-Garnet
Pyrope garnets from the West Kimberley Province were sourced from 95 to 215 km depth, with most samples coming from 140 to 190 km (Figure 4).The equilibration T ranges from ∼720 to 1335°C.Garnets show variation in major element compositions with lithospheric depth, notably TiO 2 and Cr 2 O 3 (Figure 9).The strongest enrichments in TiO 2 are recorded at the base of the lithosphere (Figure 9).This enrichment is accompanied by higher Na 2 O (see Supporting Information S1 and S2).The Mg# ranges between 0.71 and 0.87 and Cr 2 O 3 contents between 1 and 11 wt %, with the highest values observed within the deep CLM (Figure 9).The calculated Mg# of olivine coexisting with garnet (Mg# olv ) ranged between 0.82 and 0.94, with most samples recording values of 0.90-0.92(Figure 9).The range in Mg# olv agrees well with values determined from micro xenoliths and olivine xenocrysts from the West Kimberley Province (Jaques & Foley, 2018).The calculated Mg# olv across the Kimberley Craton decreases slightly with depth, with the lowest values occurring along the LAB (Figure 9).The Mg# olv also decreases with increasing TiO 2 contents in garnet.Trace element enrichments were recorded for most HFSE and REEs (Figure 9).Enrichment in P (phosphorous) was observed within the garnets from the deep CLM and along the base of the LAB which may be associated with similar enrichments in Na 2 O.The concentration of Sc, V, Sr, Zr, and Nb were typically highest at the base of the lithosphere, with their concentrations Changes in MREE and HREE concentration are less pronounced in comparison to LREEs (Figure 10).Pyropes that equilibrated at shallower depths are more depleted in LREE than those from the deep CLM (Figure 10).The highest Zr/Y values occur within the deep CLM.
Pressure-temperature estimates for garnets from the North Kimberley Province range between 120 and 240 km and 900 and 1500°C (Figure 4).Most of the garnets equilibrated within the deep CLM (>150 km) with the highest temperature garnets (>1450℃) found in the Skerring pipe.The concentration of Cr 2 O 3 and TiO 2 increases with lithospheric depth with the highest concentrations near the base of the lithosphere (Figure 9).The highest Cr 2 O 3 concentrations (up to 15 wt %) are found in the G10 garnets of the Bulgurri kimberlite whereas the garnets with the highest TiO 2 are found in the Skerring pipe garnets (Figure 9).The Mg# ranges from 0.78 to 0.88 and increases with lithospheric depth with the highest values occurring within the deep CLM (Supporting Information S1 and S2).The Skerring pipe is an exception and contains several Fe-rich garnets (Mg# ∼0.79-0.83)equilibrated at depths greater than 200 km (Supporting Information S1 and S2).The calculated Mg# olv of olivine's coexisting with garnet ranges between 0.85 and 0.93, with most samples recording values of 0.91-0.92(Figure 9).The lowest values for Mg# olv are recorded for high-Ti garnets.Garnets that equilibrated within the deep CLM recorded enrichments in Sc, Sr, Zr, and Nb, and depletions in Y with the higher Ti garnets from Skerring being the exception (Figure 9).The PM-normalized REE patterns of garnets ranges from LREE-depleted to flat, and weakly enriched (Figure 10).The range in REE abundances is illustrated by the PM-normalized REE patterns for pyrope garnets from Bulgurri which are predominantly LREE-depleted types with a small number of HREE-depleted types (Yb N ≤ 1) of G10, H-G9 and M-G9 (∼5-15 wt% Cr 2 O 3 ) composition (Figure 10).
Pressure-temperature estimates for garnets from the East Kimberley Province ranges between 859-1330℃ and 123-239 km, although most samples equilibrated between 150 and 210 km (Figure 4).Garnets from the deep CLM record strong enrichments in TiO 2 .The concentration of FeO and Cr 2 O 3 show minimal change throughout the CLM except for slight enrichment within the deep CLM (Figure 9).Garnets have a similar range in Mg# (0.71-0.88) and Cr 2 O 3 contents (1-11 wt %) to samples from the West Kimberley CLM and are not as depleted (Mg# 0.75-0.89)and Cr-rich (up to 15 wt %) as those from the North Kimberley CLM (Figure 9).The calculated Mg# olv of olivine's coexisting with garnets ranges between 0.84 and 0.92, with most samples having values of 0.91 (Figure 9).The Mg# olv decreases with increasing depth in the mantle with the lowest values observed along the LAB.The lowest values for Mg# olv were also observed for garnets containing elevated TiO 2 .Garnets that equilibrated within the deep CLM contain higher concentrations of V, Sc, Ga, Y, and Zr than those from the shallower depths (Figure 9).Garnets from the mid-CLM contained higher concentrations of Mn and lower concentrations of light and middle REEs.Increasing REE enrichment within the deep CLM is also closely associated with depth (Figure 10).Primitive mantle normalized REE patterns for garnets from Argyle record flat, weakly enriched (relative to PM) HREE compositions and depleted LREE signatures which together define a "normal" LREE-depleted and HREE-enriched pattern for most garnets between 150 and 200 km apart from a small number with sinusoidal REE patterns (Figure 10).

Eclogites
Eclogitic garnets are found at most provinces across the Kimberley Craton, although they are most abundant beneath the West Kimberley Province (i.e., Ellendale Field) and East Kimberley Province (i.e., Argyle Lamproite).Eclogitic garnets account for approximately 10%-20% of the analyzed heavy mineral concentrate as well as diamond inclusions across the craton (Figure 5).The eclogitic garnets record a range of Mg# (0.17-0.82) and TiO 2 and Na 2 O contents (see Supporting Information S1 and S2).Most garnets do not reach the very high levels of Na 2 O (up to 0.80 wt %) and TiO 2 (up to 1.52 wt %) found in inclusions in Argyle diamonds and are closer to those found as inclusions in the Ellendale diamonds which have lower FeO, Na 2 O, and TiO 2 , and higher MgO, CaO and Sr (W.L. Griffin et al., 1988;Jaques et al., 1989a;Stachel et al., 2018).The equilibration T for coexisting garnet and clinopyroxene inclusions in Argyle eclogitic diamonds ranges from 1100 to 1350℃ based on updated Fe-Mg exchange equilibria (Sudholz et al., 2022a).Touching garnet-clinopyroxene inclusions yielded similarly high temperatures (1200-1300℃).The Argyle eclogitic diamonds are also characterized by high nitrogen aggregation temperatures of ∼1180-1320 ℃ with most lying in the range 1250-1320℃ (Bulanova et al., 2018;Stachel et al., 2018;Timmerman et al., 2019).The high T estimates for garnet-clinopyroxene pairs place their eclogitic source within the bottom 50 km of the paleogeotherm for the Kimberley Craton (Bulanova et al., 2018;Stachel et al., 2018;Timmerman et al., 2019).This assumes that the eclogitic source for diamond inclusions equilibrated along the same steady state geotherm as the peridotitic xenocryst suite.This assumption is reasonable considering that the paleogeotherms for the Kimberley Craton show very little change in lithospheric thickness and surface heat flow from the Mesoproterozoic to present (Figure 2), and peridotitic and eclogitic diamonds from Argyle have a similar range in nitrogen aggregation temperatures (Stachel et al., 2018).The eclogitic suite of diamonds at Ellendale are characterized by slightly lower T estimates (1150-1250℃) for garnet-clinopyroxene pairs, which places their eclogitic source slightly higher in the deep CLM (150-175 km).The eclogitic garnets record normal PM-normalized REE patterns with depleted LREE relative to PM and weakly enriched HREE concentrations (Figure 11).These REE patterns are similar to the normal LREE-depleted patterns of some eclogitic garnet inclusions in Argyle diamonds: others have weakly "humped" REE patterns characterized by higher LREE but lower HREE abundances than the normal types (Stachel et al., 2018) (Figure 11).The positive Eu anomalies for some eclogitic garnets have been interpreted by Stachel et al. (2018) to reflect plagioclase accumulation during the formation of the mafic crustal protolith of the host eclogite.Garnet xenocrysts from Ellendale record similar depletions in LREEs but have slightly more enriched middle and HREE contents (Figure 11).Flat HREE patterns are recorded for garnets with a weak positive, or no Eu anomaly and garnets containing a negative Eu anomaly record greater concentrations of HREEs (Figure 11).Most garnets from the Ellendale eclogite suite record a weak negative Eu anomaly.

Structure, Composition, and Lithology of the Kimberley Craton CLM
The lithospheric thickness of the Kimberley Craton ranges from 205 to 250 km (Figure 2).and S2).The composition and lithology of Cr diopsides and pyrope garnets coupled with sparse peridotite xenoliths define a layered mantle, with litho-chemical domains within the shallow (<100 km) and deep (>150 km) CLM (Figure 12).These two domains are separated by a geochemically and geophysically distinct mid-lithosphere discontinuity (MLD) that is marked by a decrease in the abundance of Cr diopside which aligns with a strong negative sP receiver function amplitude, which defines a seismically slow layer interpreted as the MLD (Fitzroy Crossing permanent station [FITZ] near Fitzroy Crossing) (Birkey et al., 2021).

Shallow CLM: Litho-Chemical Stratigraphy
The shallow Kimberley CLM (50-100 km) is comprised of low to moderate Cr 2 O 3 garnet lherzolite with a high modal abundance of Cr diopside and a low abundance of pyrope garnet (Figure 12).The shallow Kimberley CLM also comprises refractory spinel lherzolite based on xenoliths recovered from the Argyle and Ellendale lamproites (Jaques, 1986;Jaques et al., 1990;Luguet et al., 2009), the Aries kimberlite (Downes et al., 2007) and the Mad Gap Yards ultramafic lamprophyre (Downes et al., 2023).The compositions of shallow CLM Cr diopsides indicate that this layer is moderately to highly depleted but has been mildly metasomatized by LREE-rich melts or fluids, possibly related to the host kimberlite-lamproite melt and/or MLD derived melt (see below).The apparent paucity of garnets within the shallow CLM may be due to a combination of sampling bias coupled with a relatively deep spinel peridotite to garnet peridotite transition in depleted (Al-poor Cr-rich) peridotite compositions (Klemme, 2004;Ziberna et al., 2013), or the presence of depleted Cr diopside-bearing garnet-free lithologies that reflects either the earliest peridotite residues formed during partial melting, or the products of mantle metasomatism (i.e., Arndt et al., 2009;Frey et al., 1985;W. L. Griffin et al., 2003;Herzberg & Rudnick, 2012;van Achterbergh et al., 2001;Walter, 1998).van Achtenbergh et al. (2001) found that garnet-free xenoliths from the Letlhakane kimberlite (Botswana) were associated with higher modal abundances of phlogopite and Cr diopside ± Cr spinel, and lower modal garnet and orthopyroxene, which decreased with increasing degrees of modal metasomatism.Metasomatized garnet-free xenoliths at Letlhakane contained Cr diopsides with enriched concentrations of Sr, Sc, and LREEs, consistent with the Cr diopside compositions reported in Figure 7. Constraints on the primary composition of metasomes from the shallow CLM were obtained by calculating hypothetical melts in equilibrium with natural minerals, using clinopyroxene-melt distribution coefficients derived from experiments and/or demonstrably equilibrated natural assemblages.This follows the method of Aulbach et al. (2013) (see Supporting Information S1 and S2).Melts in equilibrium with the Cr diopsides from the Kimberley shallow lithosphere are similar to carbonatites (i.e., Bizimis et al., 2003) in their strong enrichment in LREEs and MREEs and weak to moderate enrichment for HREEs (Supporting Information S1 and S2).The modeled equilibrium melts differ from representative natural carbonatites in their elevated Zr-Hf-Ti compositions and slightly higher concentrations for Gd to Lu (Supporting Information S1 and S2).Modeled equilibrium kimberlite melts show a good agreement with pristine kimberlites from the Lac de Gras field (Slave Craton, Tappe et al., 2013) and Udachnaya pipe (Siberian Craton, Kamenetsky et al., 2012) (Supporting Information S1 and S2).Both compositions have elevated LREEs and a weak to moderate depletion in Zr-Hf, with the concentrations of HREEs being approximately equal to PM (Supporting Information S1 and S2).On this basis, the weak metasomatic signature preserved throughout the shallow CLM are thought to be due to melt-rock metasomatism possibly from the host magmas.

Mid-CLM: Litho-Chemical Stratigraphy
The mid-CLM (100-150 km) is characterized by a lower modal abundance of Cr diopside (Figures 4 and 12).This layer occurs beneath all provinces but is best illustrated in the West Kimberley Province.The mid-lithospheric layer records strong chemical inflections between the shallow and deep CLM (Figure 6).Chemical inflections in Cr diopsides highlight a clear disparity in the extent of LREE enrichment within the deep CLM and slightly more depleted and HREE enriched compositions within the shallow CLM.These variations are shown for La/ Yb and Sm/Er ratios in Cr diopsides (Figure 7).There are several possible interpretations of the disparity in the chemical composition between the shallow and deep CLM, and the absence of Cr diopside between 100 and 150 km (Figure 4).The base of the mid-CLM may represent a paleo-boundary, including an older LAB belonging to the shallow CLM that existed before cratonization and top-down thickening of the lithosphere (e.g., Sudholz et al., 2022b).Alternatively, or additionally, it may represent an impermeable MLD that has acted as a "sponge" for rising melts from the deep CLM or asthenosphere, thus largely shielding the shallow CLM from the extensive metasomatism that characterizes much of the deeper CLM (i.e., Aulbach et al., 2017;Selway et al., 2015).The contiguity and uniformity of the mid-CLM layer across the Kimberley Craton, and its presence beneath young (Miocene; Ellendale) and old (Mesoproterozoic; Argyle) intrusions suggests that it is a major feature of the Kimberley Craton.Further, the similarity in the depth of the Cr diopside-depleted layer with negative sP seismic receiver function amplitudes (Station FITZ, Birkey et al., 2021) suggests that it is an important geophysical discontinuity characterized by a seismically slow phase.The mid-CLM layer occurs slightly deeper than most MLDs observed globally, which are commonly marked by higher attenuation of seismic waves, higher electrical conductivity, and sharp decreases in seismic velocities, commonly between at 80 and 120 km (Aulbach et al., 2017;Ford et al., 2010;Krueger et al., 2021;Rader et al., 2015;Selway et al., 2015).Several models have been proposed for the cause of MLDs, including (a) elastically accommodated grain boundary sliding (Karato, 2012;Karato et al., 2015), (b) layers of partial melt (Kumar et al., 2012;Thybo, 2006;Thybo & Perchuć, 1997), (c) accumulation of hydrousseismically slow minerals, including pargasite and phlogopite (Aulbach et al., 2017;Kovács et al., 2017Kovács et al., , 2021;;Rader et al., 2015;Saha et al., 2021;Selway et al., 2015;Smart et al., 2021), and (d) accumulation of dense saline brines (Aulbach, 2018;Bettac et al., 2023).Evidence from experimental petrology supports a pargasite model; (a) pargasite is a stable phase at the depths inferred for most seismic MLDs (80-120 km) and (b) its stability field extends to conditions found in cratonic mantle (Mandler & Grove, 2016).The stability of pargasite is controlled by the dehydration solidus (Juriček & Keppler, 2023;Mandler & Grove, 2016;Niida & Green, 1999).Experimental studies have shown that the high-pressure range of the solidus is approximately isobaric.This feature reflects a near horizontal phase transition for pargasite between 80 and 120 km, and a significant discontinuity in the water storage capacity of peridotite within the lithospheric mantle (Green et al., 2010(Green et al., , 2014)).Although scarce, pargasite has been reported in kimberlite-hosted xenolith suites globally where large representative populations are available (see Dawson and Smith (1982) and Grégoire et al. (2005)).Its scarcity in mantle xenolith suites may relate to its low melting temperature (1000-1150℃) and susceptibility to physical and chemical weathering at the surface and during magmatic transportation.Experimental studies have shown that the modal proportions of pargasite and Cr diopside are approximately inversely correlated at mantle depths where pargasite is stable (Fumagalli et al., 2009;Juriček & Keppler, 2023;Mandler & Grove, 2016;Médard et al., 2006;Niida & Green, 1999).This is because both phases sequester many of the same elements (i.e., Si, Al, Ca, Mg) (Mandler & Grove, 2016).The inverse relationship in modal abundance suggests that the formation of pargasite is controlled in part by the H 2 O-induced breakdown of Cr diopside at the dehydration solidus.The formation of pargasite from clinopyroxene + H 2 O can be expressed through several balanced chemical reactions, including:
For the Kimberley Craton, the paucity of Cr diopside between 100 and 150 km, along with the strong-negative seismic receiver function amplitudes (i.e., Birkey et al., 2021) may correspond to a narrow layer of pargasite in peridotite.A refertilized (<88 Mg# olv ) composition for pargasite-bearing peridotite would likely be required to stabilize pargasite to depths greater than 115 km (Mandler & Grove, 2016).Alternatively, pargasite may be stabilized to very high pressures (>125 km) for peridotites with exceptionally high-water activities (a H2O > 0.5) (Juriček & Keppler, 2023).The formation of the pargasite MLD may have involved the upward percolation of H 2 O-bearing melts from the deeper CLM.As these fluids crossed the isobaric section of the dehydration solidus, they reacted with Cr diopside in fertile lherzolite to form channels of pargasite.The absence of preserved xenoliths containing pargasite hinders any constraints on the timing of the formation for this layer, however the occurrence of the layer beneath Miocene (West Kimberley Province) and Neoproterozoic (East Kimberley Province) kimberlite and lamproites pipes suggests that it may be a long-lived feature of the craton.

Deep CLM: Litho-Chemical Stratigraphy
The deep Kimberley CLM (150-250 km) is comprised of high to moderate Cr 2 O 3 garnet lherzolite, harzburgite, and eclogite (Figure 12).Covariations between Cr 2 O 3 -CaO-TiO 2 in garnets suggest that the LAB is comprised of TiO 2 -rich high-Cr 2 O 3 garnet lherzolite (Figure 12).Pyrope garnets and Cr diopsides were evenly sampled throughout the deep CLM, with the highest proportions occurring at 160-190 km (Figure 4).Chrome diopsides record elevated concentrations of TiO 2 , Cr 2 O 3 , and FeO which increase with depth toward the LAB (Figure 6).A feature of the deep CLM beneath the East and West Kimberley Provinces is the anomalous enrichments in K 2 O and BaO recorded in the Cr diopside, both the mantle xenocrysts analyzed in this study and previously reported for mantle lherzolite xenoliths (Supporting Information S1 and S2, Jaques et al., 1990;Luguet et al., 2009).Lower values for Mg# olv along the LAB (0.84-0.89) suggest that enrichment in TiO 2 and mantle refertilization were facilitated by high-T melt metasomatism.Mg# olv values of 0.90-0.93 are common for the top of the deep CLM layer (Figure 9).These values accord with range of olivine compositions (Mg# 0.90-0.928 ) found in peridotite xenoliths from the Argyle lamproite (Jaques, 1986;Jaques et al., 1990;Luguet et al., 2009) and in mantle xenocrysts in the West Kimberley lamproites (Jaques & Foley, 2018).All fall within the range of olivine compositions found in cratonic peridotite worldwide (e.g., Pearson & Wittig, 2008;Pearson et al., 2021).The garnet populations of the North Kimberley Province have high levels of Cr 2 O 3 and strongly depleted trace element signatures (Wyatt et al., 1998), suggesting that the North Kimberley Province is underlain by a strongly depleted lithosphere.
Constraints on the composition of metasomes from the deep CLM follows the method outlined in Section 6.3 (Supporting Information S1 and S2).Melts in equilibrium with Cr diopsides from the deeper Kimberley CLM agree with the bulk-rock compositions of natural representative carbonatites (i.e., Bizimis et al., 2003) in their enrichment in LREEs and MREEs and weak to moderate enrichment for HREEs (Supporting Information S1 and S2).The modeled equilibrium melts also record a weak to moderate depletion in Zr-Hf.The modeled carbonatite melts differ from natural examples in their elevated MREE compositions and higher HREEs (Gd to Lu) (Supporting Information S1 and S2).Modeled kimberlite melts agree with pristine kimberlites from the Lac de Gras field (Slave Craton, Tappe et al., 2013) and Udachnaya pipe (Siberian Craton, Kamenetsky et al., 2012) in their REE and trace element concentrations (Supporting Information S1 and S2).Both compositions are enriched in LREEs and record a weak to moderate depletion in Zr-Hf, with the concentrations of HREEs being approximately equal to PM (Supporting Information S1 and S2).The Zr-Hf depletion is slightly more pronounced in our modeled compositions in comparison to these natural examples (Supporting Information S1 and S2).Based on this evidence, we suggest that the deep CLM, was in part, enriched in LREEs during melt-rock interactions with a kimberlitic melt.The strong FeO-TiO 2 enrichments in pyrope garnets and Cr diopsides may be due to silicate-carbonatite melt metasomatism from the same kimberlite melt or reflect a separate metasomatic event relating to the upward percolation of high-T silicate melts from the asthenosphere.The latter hypothesis is supported by the lack of correlation between TiO 2 and FeO with LREEs in pyrope garnets and Cr diopsides.

Lamproite Magmatism
The deep CLM of the East and West Kimberley provinces is anomalous in its abundance of eclogitic and pyroxenitic garnet (Figure 5) and extreme enrichment in K 2 O and BaO in Cr diopsides, which are unlikely related to silicate or silicate-carbonatite melt metasomatism (Figures 6 and 7; Supporting Information S1 and S2).Substitution of significant K 2 O in Cr diopsides is limited to high pressure and temperatures (>5.0 GPa, 1200-1500℃) and K-rich bulk compositions (Harlow, 1997;Harlow & Davies, 2004;Perchuk et al., 2002).The spatial association of K-rich Cr diopsides below lamproite provinces of the West Kimberley and East Kimberley (Argyle) suggests a possible link.The lamproites at Argyle and West Kimberley are characterized by extremely potassic compositions (K 2 O/Na 2 O > 3, 2-12 wt % K 2 O) and enrichment in other LIL elements (especially Ba and LREE) as well as the HFSE elements (Ti, Zr, Hf, Nb, Jaques, 1986;Jaques et al., 1984Jaques et al., , 1989cJaques et al., , 2018)).This marked enrichment contrasts with the very low abundances of HREEs, Sc, and V (and Al and Na) in the lamproites, which are close to, or more depleted than PM and indicate depleted garnet-bearing mantle peridotite in the source region.The lamproites are thought to have formed by very small degrees of partial melting of a refractory garnet-poor lherzolite (depleted in Al, Ca, Na, Sc, V, HREEs) that was depleted by melting in the garnet stability field and then metasomatized by melts with high abundances LIL and other incompatible elements (Jaques et al., 1984(Jaques et al., , 1989a(Jaques et al., , 2018;;Tainton & McKenzie, 1994).This metasomatized source is inferred to include phlogopite (high K 2 O/Al 2 O 3 ∼1) and potentially other K-bearing phases to generate the high K 2 O/Al 2 O 3 ratios of the lamproites (Foley, 1992).Both peridotitic and eclogitic inclusions in Argyle diamonds have very high K 2 O contents (Jaques et al., 1989b;Stachel et al., 2018) as do many of the Cr diopsides in peridotite xenoliths from Argyle (Jaques et al., 1990(Jaques et al., , 2018;;Luguet et al., 2009).We suggest that the K 2 O-rich Cr diopsides in the deep mantle beneath the Argyle and West Kimberley lamproites may be remnants of this strongly metasomatized CLM that gave rise to or, at least, contributed to the lamproite magmas.The high 87 Sr/ 86 Sr and low 143 Nd/ 144 Nd ratios of the lamproites indicate they were derived from CLM that has undergone long-term enrichment by metasomatism or at least include a significant component of an ancient (≥2,000 Myr) enriched (high Nd/Sm, Rb/Sr) mantle source(s) (Fraser et al., 1985;Jaques et al., 1989bJaques et al., , 2018;;McCulloch et al., 1983;Nelson et al., 1986).Subducted sediment derived from the continental crust has been suggested as the source of the LIL geochemical enrichment of the lamproites based on their low K/Rb, K/Ba, Ce/Pb, and high Th/U; their high 87 Sr/ 86 Sr and unradiogenic Nd isotopic compositions; and, for the West Kimberley lamproites, their unusual Pb isotopic compositions with high 207 Pb/ 204 Pb and low 206 Pb/ 204 Pb ratios (Nelson, 1992;Nelson et al., 1986).Subduction of oceanic crust at the margin of the craton is inferred to be the source of the eclogite protoliths and crustal (biogenic) carbon that contributed to the isotopically light carbon isotopes of the Argyle eclogitic diamond suite (Jaques et al., 1989b(Jaques et al., , 2018;;Stachel et al., 2018).

Formation and Evolution of the Kimberley Craton CLM
The age of the CLM underlying the Kimberley Craton has been constrained to the Mesoarchean to Paleoproterozoic (2,200 Ma) from Re-Os dating of peridotitic xenoliths and sulphide inclusions in diamond (Graham et al., 1999;Luguet et al., 2009;Smit et al., 2010).A similar age was suggested for the lithosphere under the North Kimberley region based on Os isotopic data for chromite from the Seppelt kimberlite (Graham et al., 1999).Because most Re-Os T RD ages were determined for high-P peridotite xenoliths (>150 km), this age provides an estimate for the deep CLM.The absence of Re-Os data for low-P peridotite xenoliths (<100 km) precludes age estimates for the shallow CLM, however, it is assumed to be at least Neoarchean in age, consistent with common presence of Neoarchean (∼2,500 Ma) detrital zircons, thought to be derived from unexposed local juvenile mantle-derived crustal sources, in overlying sedimentary sequences (Hollis et al., 2014;Maidment et al., 2022).The Re-Os T RD ages for peridotite xenoliths, and therefore the age of cratonic mantle formation, cannot be directly correlated with any known crustal rocks or tectono-thermal events within the Kimberley Craton.However, they do correlate with the timing of widespread granite magmatism recorded across Archean cratons globally (including for neighboring Yilgarn and Pilbara Cratons), which is interpreted to reflect the onset of modern-style plate tectonics and craton stabilization (Cawood et al., 2018).
The disparity in the chemical compositions of Cr diopsides between the shallow and deep CLM domains suggests that they may have had different origins or least different metasomatic histories.We suggest that the shallow CLM represents an early nucleus of the Kimberley Craton.The chemical compositions of Cr diopsides, and the inferred garnet-free lithologies from the shallow CLM provides clues for two possible models of formation for this domain.(a) The shallow CLM layer and the MLD layer, prior to modification by hydrous melts, may reflect a refertilized depleted domain that formed as the residues of high degrees of partial melting within the bounds of the spinel stability field and garnet stability field associated with early (Neo to Mesoarchean) basaltic magmatism.The current suite of Cr diopsides may have been added to the shallow CLM during a later metasomatic-refertilization event that involved the addition of Al 2 O 3 and CaO into the initially depleted peridotite.(b) An alternative, albeit simpler explanation for the shallow CLM layer is that the abundant Cr diopsides and paucity of garnet is due to the high-P stabilization of spinel in refractory peridotite at the expense of garnet during partial melting to form residual spinel lherzolite.This interpretation is challenged by garnet-spinel equilibrium modeling by Ziberna et al. (2013) which showed that changes in the bulk rock concentration of Cr 2 O 3 and Al 2 O 3 in peridotite can widen the stability field of spinel to higher pressures but does not significantly affect the stability range of garnet (specifically the depth of the garnet-in reaction).Nevertheless, the diversity in mineral chemistry within the shallow SCLM across the kimberlite and lamproite provinces suggests that the shallow CLM may have formed by the amalgamation of different lithospheric domains.This interpretation accords with the view that the Kimberley Craton, at surface, comprises a series of linear, north-north-easterly trending Archean to Paleoproterozoic terranes (Gunn & Meixner, 1998;Tyler et al., 1999).
The base of the shallow Kimberley CLM layer occurs at a similar depth to the dehydration solidus of hydrous-fertile peridotite (<88 Mg# olv ) (Juriček & Keppler, 2023;Mandler & Grove, 2016;Niida & Green, 1999), which may relate to the accumulation of pargasite through metasomatic replacement of Cr diopside (see Section 6.3).A pargasite mid CLM layer might have formed when hydrous metasomatic fluids or hydrous melts from the asthenosphere at the base of the CLM percolated upwards and crossed the isobaric section of the dehydration solidus between 100 and 130 km (Green et al., 2010(Green et al., , 2014)).The chemical disparity between the shallow and deep CLM domains raises the possibility that the shallow CLM may have formed prior to the addition of the deeper CLM layer.The absence of preserved xenoliths from the mid CLM layer precludes any constraints on the timing of xenoliths that enrichment in LREE also occurred at this time but noted that this represents a minimum age.These authors also suggested a metasomatic enrichment event at ∼1,500 (±150) Ma based on Os isotopic data for a picroilmenite xenocryst from the Maude Creek kimberlite.However, if the terranes of the Lamboo Province in the Halls Creek orogen are para-autochthonous then subduction and mantle enrichment could pre-date the Halls Creek Orogeny (Maidment et al., 2022).The old Re-Os T RD ages for the Argyle peridotite xenoliths suggest that although accretionary events likely contributed toward some of the thickening of the Kimberley Craton CLM, its thickness was likely largely acquired during the Archean or earliest Proterozoic.In the absence of known accretionary events older than 1,800-1,900 Ma for the Kimberley Craton, we infer that the depleted compositions for pyrope garnets and elevated Mg# olv (∼0.90-0.93)for the deep CLM reflects the residues of ancient partial melting events.Slight bottom-up increases in the Mg# olv throughout the deep CLM suggests that this layer might have formed through polybaric melt extraction (Arndt et al., 2009;Herzberg, 2004).The range in calculated Mg# olv (∼0.90-0.93)implies that 20%-40% partial melt extraction was likely required in order to generate the observed garnet-olivine compositions.A similar model involving melting over a large depth range, probably starting in the garnet stability field but ending at shallow depths, followed by thickening either by simple shortening during terrane collision or via subduction was proposed for formation of the mantle beneath the Argyle lamproite (Jaques et al., 2018).The range in Mg# olv (∼0.90-0.93)may also have been influenced by the addition of FeO-TiO 2 during high-T metasomatism from silicate-carbonatite melts (see Section 6.4).The range in depleted and metasomatized-refertilized compositions for pyrope garnets and Cr diopsides, along with the presence of eclogitic domains between 150 and 250 km, may be explained by a hybrid model for the formation of the deep CLM that involved polybaric melt extraction during the Archean and subsequent thickening during arc accretion along the margin of the Kimberley Craton during the Paleoproterozoic.The Paleoproterozoic Hooper Orogeny (1,860) and Halls Creek Orogeny (1,805-1,835 Ma) are the oldest known orogenic events recorded across the Kimberley Craton.Both orogenic events have been inferred to mark the amalgamation of the Kimberley Craton within the larger North Australian Craton during the formation of the Nuna Supercontinent (Betts et al., 2016).Oceanic crust and associated overlying sediments that were subducted at the craton margin during these events likely enriched the deep CLM in these regions in volatiles and incompatible elements (K, Ba, Pb).Formation of the Argyle eclogitic suite diamonds at ∼1,580 Ma (Richardson, 1986) is attributed to subduction of oceanic crust based on the chemistry of the mineral inclusions and their carbon (δ 13 C ∼11%), oxygen and noble gas compositions (Jaques et al., 1989b;Stachel et al., 2018;Timmerman et al., 2019).Additional episodes of growth of eclogitic diamonds associated with younger orogenies were proposed by Smit et al. (2010) for the Ellendale diamonds based on Re-Os dating of sulphide inclusions.

Timing of Kimberlite and Lamproite Magmatism
The ∼1,200 Ma Argyle lamproite and associated Lissadell Road and Seagull's dykes represent a small-scale mantle melting event involving very limited and localized melt production.Emplacement of the Argyle lamproite into the Paleoproterozoic Halls Creek Orogen is inferred to have occurred during a period of extension associated with prominent large-scale NNE-trending strike-slip faulting (Rayner et al., 2018).The ∼1,200 Ma age of the Argyle pipe coincides with a sharp bend in the apparent polar wander path for Australia at this time (Wingate & Evans, 2003) and the proposed separation of Australia from the Nuna supercontinent (Kirscher et al., 2021).It also overlaps the well-known 1,100-1,200 Ma peak in global kimberlite, lamproite and related alkaline magmatism attributed to the reorganization of tectonic plates (Jelsma et al., 2009;Pandey & Rao, 2020;Tappe et al., 2018).Kimberlite magmatism during the Neoproterozoic (∼800-850 Ma) was focused on the North Kimberley Province, and to a lesser extent, the East and Central Kimberley Provinces.This magmatism occurred during the early stages of the breakup of the Rodinia supercontinent (Downes et al., 2016(Downes et al., , 2023;;Jaques & Milligan, 2004), and coincided with several small-to medium-scale volcanic events elsewhere on the Australian continent, including the Gairdner large igneous province (LIP) (827 Ma, Wingate et al., 1998), Mundine Well Dyke Swarm (755 Ma, Li et al., 2006;Wingate & Giddings, 2000) and Boucaut Magmatic Event (788 Ma; Armistead et al., 2021), as well as aillikite and related magmatism within the Gibson Desert (806 Ma, Webb Kimberlite Province) (Western Australia).Formation of these LIPs and related magmas has been attributed to a "superplume" event which led to continental rifting and the breakup of Rodinia (Li et al., 2006).The Neoproterozoic (∼800-850 Ma) kimberlites and lamprophyres of the Kimberley Craton appear to have formed in response to early stage plume impingement on the metasomatized roots of the North Australian Craton during the incipient stages of the breakup of Rodinia.The (limited) isotopic data available suggest that the magmas were derived from the asthenosphere (Downes et al., 2023;Edwards et al., 1992;Fielding & Jaques, 1986;Graham et al., 1999) in contrast to the significant input of enriched CLM seen in the isotopic compositions of the Argyle and West Kimberley lamproites (e.g., Jaques et al., 2018 and refs therein).The high equilibration temperatures of the sub-calcic Cr diopside xenocrysts and abundant picroilmenite in the Skerring kimberlite likely reflects a refertilization event involving plume-derived magma at or close to the eruption age of Skerring (Taylor et al., 1999).The most recent (∼20 Ma) period of alkaline-ultramafic magmatism within the West Kimberley Craton involved the eruption of more than 150 lamproite pipes, sills and plugs across a 200 km 2 area.Unlike neighboring kimberlite and lamproite provinces, the Miocene lamproite volcanism does not coincide with a major period of alkaline magmatism globally.This suggests that it was a localized event not related to a mantle plume.Magmatism is thought to have involved the melting of veined CLM perhaps initiated by the addition of a small heat or melt/fluid flux from the asthenosphere.Phillips et al. (2022) attributed the lamproite magmatism to decompression melting on uplift and/ or small-scale mantle convection near the craton margin with magmatism exploiting major trans-lithospheric structures.The lamproite magmatism occurred during a rapid burst of accelerated northward plate motion of the Australian Plate with most of the lamproites erupting during the second of two phases of rapid northward movement of the Australian Plate between 30 to 34 Ma and 16 to 23 Ma (Cohen et al., 2013).

Seismic Tomography
Seismic tomographic models of the Australian lithosphere show that the Kimberley Craton is underlain by a thick (up to 250 km) lithospheric root that thins to the east and south and extends beyond the Proterozoic orogenic belts at the craton margin (de Laat et al., 2023;Fishwick & Rawlinson, 2012;Fishwick et al., 2005;Kennett, 2003;Magrini et al., 2023).The most recent S-wave velocity tomography models of Australia (Aus22, de Laat et al., 2023;Magrini et al., 2023) propose a thin lithosphere beneath the West Kimberley Province.The occurrence of thin lithosphere, which differs from earlier models (i.e., Fishwick & Rawlinson, 2012), may relate to the omission of the FITZ from the model of de Laat et al. ( 2023), or anomalous thermal architecture and additional slow seismic phases relating to an MLD (i.e., Birkey et al., 2021) within the shallow CLM of the West Kimberley Province (i.e., Figure 2).The occurrence of thin lithosphere (<200 km) beneath the West Kimberley Province, is not supported by our updated FITPLOT paleogeotherms for diamondiferous Miocene (∼20 Ma) lamproites from the Ellendale, Calwynyardah, Walgidee Hill and Laymans Bore lamproites (Figure 2).The FITPLOT paleogeotherm LAB estimates range between 225 and 245 km for the West Kimberley Province (Figure 2).Similar FITPLOT paleogeotherm depth to LAB estimates for Miocene to Mesoproterozoic lamproites and kimberlites from adjacent provinces within the Central, Eastern and Northern Provinces of the Kimberley Craton suggest that the thick cratonic lithosphere of the Kimberley Craton has remained unchanged since at least 1,000 Ma and that thermal erosion of the CLM is unlikely.The thick cratonic lithosphere beneath the West Kimberley Province is further attested by receiver functions calculated from the FITZ permanent station.Results from Birkey et al. (2021) define a prominent slow seismic phase at 240-260 km, which aligns with the top of our petrological LAB.The agreement between both methods provides evidence that the geodynamic interpretations of de Laat et al. (2023) (see Section 5.4 of de Laat et al., 2023) for thin un-cratonized lithosphere for the West Kimberley Province involving thermal erosion by a cryptic mantle plume are incorrect and that thick cratonic lithosphere of the Kimberley Craton extends beneath both the Halls Creek and Wunaamin Miliwundi orogens, as least as far as the Walgidee Hills lamproite.

Conclusion
The lithology, geochemistry, and architecture of the CLM underlying the Kimberley Craton have been constrained using PT estimates and mineral compositions for >5,000 This layer was also formed as a residue from peridotite melting and was either accreted onto the shallow CLM or formed contemporaneously with the shallow CLM by polybaric melting during the Late Archean.The Cr diopside and garnet xenocrysts record enrichment in LREEs, K, Ba, and P as well as Ti and other elements which is attributed to metasomatism from asthenospheric melts and by Paleoproterozoic subduction (K.Ba) events at the craton edge that were also responsible for eclogite deposition and diamond formation.ZS acknowledges financial and research support from the Exploring for the Future program of Geoscience Australia.Several of the garnet and diopside suites from heavy mineral concentrate were prepared at Diatech Mineral Services, Perth.Garnet and diopside analyses were obtained at the Centre for Advanced Microscopy, a division of Microscopy Australia, an organization that is funded by the Australian National University and the state and federal government of Australia.Jeff Chen and Corinne Frigo are kindly thanked for their assistance with EPMA analyses.Brett Knowles and Monika Misztela are kindly thanked for their assistance with LA-ICP-MS analyses.Bill Griffin (Macquarie) and Bruce Wyatt are thanked for providing published analyses of North Australian garnets.India Bore Diamond Holdings Pty Ltd. are thanked for providing some of the heavy mineral concentrate used in this study.Sonja Aulbach and Roberta Rudnick are kindly thanked for their comprehensive reviews which helped to strengthen the manuscript and iron-out several inconsistencies.Paul Asimow is kindly acknowledged for editorial handling.Michael Doublier (Geoscience Australia) and Evgeniy Bastrakov (Geoscience Australia) are thanked for providing constructive reviews on an earlier version of the manuscript.KC and MH publish with permission of the CEO of Geoscience Australia.Geoscience Australia eCat number 147708.Open access publishing facilitated by Australian National University, as part of the Wiley -Australian National University agreement via the Council of Australian University Librarians.

Figure 1 .
Figure 1.Simplified geological map of the Kimberley Craton of Western Australia.Open squares are towns.
Fishwick and Rawlinson (2012) and Hoggard et al. (2020) but differ by an estimated 50-100 km from the tomography 10.1029/2023GC011040 9 of 32 models of de Laat et al. (

Figure 2 .
Figure2.FITPLOT and conductive paleogeotherms for the Kimberley Craton.Paleogeotherms were calculated using the method described byMather et al. (2011).See Supporting Information S1 and S2 for conductive model geotherms for the Kimberley Craton (afterHasterok and Chapman (2011)).A fixed crust-mantle boundary of 40 km and a mantle isentrope of 1330℃ were used for all models.

Figure 4 .
Figure 4. Histogram of (a) pyrope garnet and (b) Cr diopside equilibration depths for kimberlite and lamproite intrusions across the Kimberley Craton.BIN size is 10 km.The color code follows Figure 3. Count (n) on x-axis is the number of (a) garnet and (b) Cr diopside.This figure shows that the shallow continental lithospheric mantle (CLM) (<100 km) comprises Cr diopside rich lithologies that are depleted in garnet.The mid CLM (100-150 km) is depleted in Cr diopside and pyrope garnet.The deep CLM (>150 km) contains peridotite lithologies rich in Cr diopside and pyrope garnet.

Figure 5 .
Figure 5. Pie diagrams for garnet type for kimberlite and lamproite provinces of the Kimberley Craton.Garnet classification from Figure 3. See Section 4.2 and Grütter et al. (2004) for discussion of garnet classification.This figure highlights a prominent eclogitic (G3) mantle component for the East Kimberley Province.The West and North Kimberley Provinces comprise moderate to high Cr garnet peridotite components (M-G9, H-G9, and G10) along with Ti-metasomatized litho-types (G11 and G1) G1 garnets are particularly prominent in the Skerring and Maude Creek kimberlites.G10 (low Ca garnets) are prominent in the Bulgurri kimberlite and to a lesser extent the Maude Creek and Seppelt kimberlites and Ellendale lamproites.

Figure 6 .
Figure 6.Major oxide chemical stratigraphies for Cr diopside xenocrysts from kimberlites, lamproites, and ultramafic lamprophyres of the Kimberley Craton.This figure highlights the vertical change in the major and minor oxide composition of Cr diopsides from the Kimberley Craton.Note the significant compositional changes between the shallow and deep lithosphere sections.

Figure 7 .
Figure 7. Trace element chemical stratigraphies for Cr diopside xenocrysts from kimberlites, lamproites, and ultramafic lamprophyres of the Kimberley Craton.This figure highlights the vertical change in the trace element composition of Cr diopsides from the Kimberley Craton.

Figure 8 .
Figure 8. Depth slices for average primitive mantle (PM) normalized rare earth element patterns in Cr diopside from the West Kimberley lamproites.PM values from McDonough and Sun (1995).The color code follows Figure 3.

Figure 9 .
Figure 9. Major-minor oxide and trace element chemical stratigraphies for pyrope garnet xenocrysts from kimberlites, lamproites, and ultramafic lamprophyres of the Kimberley Craton.The color code follows Figure 6.This figure highlights the vertical change in the major and minor oxide, and trace element composition of pyrope garnets from the Kimberley Craton.

Figure 10 .
Figure 10.Depth slices for average primitive mantle (PM) normalized rare earth element patterns in pyrope garnet xenocrysts from kimberlites and lamproites lamprophyres of the Kimberley Craton.PM values from McDonough and Sun (1995).

Figure 12 .
Figure 12.Simplified litho-chemical stratigraphic column for the Kimberley Craton.Shallow continental lithospheric mantle (CLM) (<100 km): moderately depleted and weakly melt-metasomatized peridotite that comprises common Cr diopside and depleted pyrope garnet.Mid CLM (100-150 km): refertilized pargasite-bearing (inferred) lherzolite depleted in Cr diopside.The pargasite is inferred to have been added to the Mid CLM through the H 2 O induced break-down of Cr diopside at the dehydration solidus of a refertilized (<88 Mg# olv ) peridotite.Deep CLM (150-200 km): Eclogite lenses throughout the deep CLM which may reflect remnants of the Proterozoic subduction of oceanic crust.The orientation and depth distribution of eclogite is inferred.Majority of deep CLM comprises depleted and refertilized peridotite.These rocks have been modified by high-T FeO-TiO 2 melts from the asthenosphere (silicate-carbonatite melts) and potassic enriched fluids associated with subduction.
the formation for this layer: it may have formed prior to the addition of the deeper Kimberley CLM in which case the base of the mid CLM might represent a paleo LAB.Alternatively, both the shallow and deep CLM may have formed at the same time and the MLD layer may have formed during one of several metasomatic episodes associated with the addition of HSFE and LREEs into the deeper CLM.The deep CLM domain beneath the Kimberley Craton extends from 150 km to the base of the lithosphere (∼250 km).The compositions of garnet and Cr diopside indicate that this layer is the most intensely metasomatized section of the CLM (see Section 6).At least three metasomatic events are recorded throughout the deep CLM: (a) K 2 O-BaO metasomatism beneath the Paleoproterozoic Halls Creek and Wunaamin Miliwundi orogens of the East and West Kimberley Provinces from melts or K-Ba-rich brines derived from subducted continental sediments, (b) LREE metasomatism from a kimberlite-lamproite melt, and (c) High-T TiO 2 -FeO metasomatism from a silicate-carbonatite melts sourced from the asthenosphere.The Re-Os T RD ages for high-P peridotite xenoliths indicate that the deep Kimberley CLM formed approximately 1 billion years before the Paleoproterozoic arc accretion events at ∼1,900-1,800 Myr that have been inferred to be responsible for eclogite deposition and diamond formation.Graham et al. (1999) suggested based on depleted mantle model ages for Argyle peridotite garnet and Cr diopside mantle xenocrysts from 25 kimberlite and lamproite intrusions.Paleogeotherms define a lithospheric thickness of 200-250 km and surface heat flow of 37-40 mW/m 2 for most of the kimberlite and lamproite provinces.Similar values for Miocene and Mesoproterozoic intrusions suggest that the lithospheric architecture has not changed over the past 1,000 Ma.The chemical compositions of Cr diopsides and garnet define a stacked lithosphere with distinct geochemical domains in the shallow (<100 km) and deep (>150 km) CLM, separated by a Cr diopside-depleted and seismically slow MLD (100-150 km).The shallow CLM is comprised of spinel lherzolite and low to moderate Cr 2 O 3 garnet lherzolite with a high modal abundance of Cr diopside.These phases record evidence of both variable degrees of depletion by partial melting and basaltic melt extraction and subsequent enrichment in LREE and other elements.This layer is interpreted as an earlier nucleus of the Kimberley Craton.The base of this layer is now the site of a geochemically and seismically distinct MLD depleted in Cr diopside and garnet which is inferred to have formed by reaction of hydrous melts with peridotite on crossing the dehydration solidus.The deep CLM is comprised of high to moderate Cr 2 O 3 lherzolite, lesser harzburgite, and eclogite.

Table 1
, and This work Sample Location, Intrusion Age, and the Number of Analyzed Garnet and Cr Diopside Grains Used in the Study Hasterok and Chapman (2011)orded across the entire craton, although the deepest estimates occur within the West and North Kimberley Provinces.Based on the occurrence of thick (up to 250 km) lithosphere and the presence of depleted garnet compositions, we suggest the North Kimberley Province likely reflects the deepest and most refractory peridotite section of the Kimberley Craton.Comparable LAB estimates beneath Mesoproterozoic, Neoproterozoic, and Miocene pipes indicate that the lithospheric thickness and thermal state of the CLM has remained stable over the past 1,000 Ma.The CLM underlying the Kimberley Craton is thermally stable, with most samples from the deep CLM (>150 km) plotting between the 35 and 40 mW/m 2 model conductive paleogeotherms ofHasterok and Chapman (2011)(Supporting Information S1