https://doi.org/10.1130/G45180.1 Manuscript received 21 May 2018 Revised manuscript received 2 October 2018 Manuscript accepted 8 October 2018 © 2018 Geological Society of America. For permission to copy, contact [email protected]. Published online 23 October 2018 Oceanic origin of continental mantle lithosphere Andrea Servali* and Jun Korenaga Department of Geology and Geophysics, Yale University, PO Box 208109, New Haven, Connecticut 06520-8109, USA ABSTRACT between the emerged trend of mantle depletion We present a global compilation of major element, as well as Re-Os isotope, data on and the secular cooling of the upper mantle. mantle xenoliths from continental lithosphere to constrain the secular evolution of mantle depletion since the early Archean. Whereas a temporal dichotomy in the degree of mantle MANTLE XENOLITHS depletion has long been recognized in previous regional studies of mantle xenoliths, this global The chemical composition of continental compilation reveals, for the first time, a smooth secular trend in mantle depletion, which mantle lithosphere is highly variable around is in remarkable agreement with what is expected from the secular cooling of the ambient the globe. Whereas Archean and younger cra- mantle as inferred from the petrology of non-arc basalts. Depleted mantle now composing tons are generally underlain by highly depleted continental lithosphere is likely to have been originally formed beneath mid-ocean ridges or mantle and less-depleted mantle, respectively, similar spreading environments, and a greater degree of depletion in the past can be seen as each craton has been affected by its own tectonic a corollary of the secular cooling of the mantle. history. Commonly, the degree of complexity observed in surface geology is coupled with an INTRODUCTION Aulbach et al., 2011). Additional depletion in arc equivalent amount of deformation in the man- A plethora of evidence from tomographic settings has also been suggested (e.g., Pearson tle (e.g., Carlson et al., 2005). As a result, the imaging and petrological observations on man- and Wittig, 2008). Resolving the origin of conti- mantle lithosphere preserves changes in major tle xenoliths corroborates the existence of ther- nental lithosphere has been difficult because the and minor element compositions correlated to mally and chemically distinct continental mantle occurrence of mantle xenoliths is spatially lim- tectonic processes, such as metasomatism, that lithosphere underlying most Archean and Pro- ited and the geological record of old continental affect interpretation of data. To account for terozoic exposed landmasses (e.g., Carlson et al., crust is generally complex. In this study, we aim possible modifications of the original Mg# and 2005). Petrological analyses of cratonic mantle to test the oceanic ridge hypothesis on the basis Re-Os isotope system, it is important to focus xenoliths from Archean and Proterozoic terranes of a global compilation of mantle xenolith data. on geological areas that have been thoroughly indicate a mantle predominantly composed of One attractive feature of this hypothesis is that studied, with a statistically sufficient number highly depleted peridotites (e.g., Boyd, 1989; it makes a definitive prediction for the secular of whole-rock analyses (Rudnick and Walker, Pearson and Wittig, 2008). The degree of deple- compositional trend of mantle depletion (Herz- 2009). Except for some Mesozoic and Paleozoic tion is commonly described by Mg#, which is berg and Rudnick, 2012) against which xenolith peridotite xenoliths retrieved in alkaline basalts molar Mg / (Mg + Fe) × 100, and a higher Mg# data can be compared. Whereas a contrast in from the China blocks, mantle xenoliths in our results from a higher degree of melting. The the degree of depletion between Archean and compilation were collected in kimberlite pipes average Mg# of cratonic xenoliths is ~92, reach- Proterozoic xenoliths has long been known, as that erupted between the Proterozoic and the ing as high as 94 for some Archean samples, noted above, a detailed picture of how mantle Mesozoic (e.g., Irvine et al. 2003; Pearson et al., whereas typical fertile peridotites in the present depletion has changed through time is yet to 2007; Wittig et al., 2010; Liu et al., 2012). Good asthenospheric mantle have a Mg# ~88–89. The be established. To this end, we assemble major age correlation is generally observed between Mg# of mantle xenoliths has long been known element composition (i.e., MgO and FeO), as model ages of mantle xenoliths and ages of to exhibit a temporal dichotomy; samples col- well as Re-Os isotope data, for mantle xenoliths crustal rocks, suggesting a coupling between lected from Archean cratons are characterized found on major cratons. crustal and mantle lithosphere (Rollinson, 2010); by higher levels of depletion compared to those In what follows, we first provide essential notable exceptions to this general trend are the from Proterozoic and Phanerozoic provinces traits of the cratons and xenoliths we consider Sino-Korean craton and the South China block. (Boyd, 1989; Kelemen et al., 1998; Griffin et in this study. Such information aids in interpre- The data set includes samples from the Churchill al., 2008). tation of the wide range of ages and chemical province (Canada); Kaapvaal (southern Africa), The existence of highly depleted mantle has compositions present in our global compilation. Tanzania, North and South China, North Atlantic led to two popular hypotheses for the origin of We then describe our strategy for data compila- (Greenland), Slave (Canada), Siberian, and Kare- continental mantle lithosphere: one involving tion to extract the temporal evolution of mantle lian (Finland and Russia) cratons; and Wyoming oceanic ridges (e.g., Herzberg, 2004; Rollinson, depletion. Finally, we discuss a possible origin province (USA and Canada) (sample locations, 2010; Pearson and Wittig, 2014) and the other of continental mantle lithosphere via mid-ocean source references, and data are provided in the involving mantle plumes (e.g., Herzberg, 1999; ridge processes, on the basis of a correlation GSA Data Repository1, along with a summary of 1 GSA Data Repository item 2018397, supplementary information on surface geology of cratons, supplementary figures, and Table DR1(data and sources for all cratons), is available online at http://www.geosociety.org/datarepository/2018/ or on request from [email protected]. *E-mail: [email protected] CITATION: Servali, A., and Korenaga, J., 2018, Oceanic origin of continental mantle lithosphere: Geology, v. 46, p. 1047–1050, https://doi.org/10.1130/G45180.1 Geological Society of America | GEOLOGY | Volume 46 | Number 12 | www.gsapubs.org 1047 Downloaded from https://pubs.geoscienceworld.org/gsa/geology/article-pdf/46/12/1047/4572943/1047.pdf by Jun Korenaga on 01 December 2018 95 95 notable surface geology characteristics for each On Kaapvaal craton of these cratons). Off Kaapvaal craton ABNorth China block Tanzania craton South China block 94 Ur = 0.2 94 Ur = 0.2 METHODS AND RESULTS During the partial melting of the mantle, 93 93 Fe and Re are partitioned into the melt phase Ur = 0.25 Ur = 0.25 more preferentially than Mg and Os, respec- 92 92 tively. Such a contrast in element partitioning is Mg# Mg# central to quantifying the degree of melting and 91 Ur = 0.3 91 Ur = 0.3 its timing, the former by Mg-Fe partitioning and the latter by Re-Os model ages. For Mg-Fe par- 90 90 titioning, we consider both olivine and whole- rock Mg#. Re-Os model ages rely on sulfides 89 89 that are susceptible to secondary processes and thus are not expected to provide accurate ages; 88 88 01234 01234 numerous pre-, syn-, and post-eruption chemi- Age (Ga) Age (Ga) 95 95 cal interactions between melt and residue com- Churchill prov. monly alter Re and Os concentrations, thereby On North Atlantic craton C Karelian craton D Off North Atlantic craton Siberian craton perturbing estimates on the timing of mantle 94 Ur = 0.2 94 Slave craton Ur = 0.2 depletion (Pearson and Wittig, 2008; Liu et al., Wyoming prov. 2010). Such model ages, however, represent the 93 93 only effective method to constrain the timing of Ur = 0.25 Ur = 0.25 mantle depletion (Rudnick and Walker, 2009). 92 92 Fortunately, two different types of model ages, Mg# Mg# 91 Ur = 0.3 91 Ur = 0.3 Re depletion (TRD) and mantle (TMA) model ages, have been devised (Walker et al., 1989), which allow discussion of the reliability of age esti- 90 90 mates (Carlson et al., 2005; Pearson and Wit- tig, 2008). 89 89 The majority of the published Re-Os isotope 88 88 data for mantle xenoliths are based on whole- 01234 01234 rock analysis. Unlike sulfide analysis, whole- Age (Ga) Age (Ga) rock analysis does not allow probing of the pos- Figure 1. Covariation of whole-rock Mg# [molar Mg / (Mg + Fe) × 100] and model ages for sibility of secondary metasomatism, but model mantle xenoliths from Kaapvaal and Tanzania cratons, Africa (A), China cratons (B), the North ages based on whole-rock analyses generally Atlantic craton (C), and other major cratons (see text for locations (D). (prov.—province) For the majority of data, both Re depletion (T ) and mantle (T ) model ages are shown (with T reflect the age of crystalline basement or later RD MA MA being older). Solid symbols denote TRD model ages corrected for eruption contamination, tectonic deformation (Griffin et al., 2004; Pear- whereas open symbols denote those uncorrected. Dark blue symbols connected with line son and Wittig, 2014). We thus adopt whole- denote data with TMA – TRD < 0.2 b.y. Light blue symbols are used for data with TMA – TRD < 1 b.y. Symbols connected by dashed line are used for data uncorrected for eruption contamination. rock TRD and TMA model ages as the lower and Gray symbols represent T ages for data with T >1 b.y.