save for Davis et al. (2017), were obtained using spinel standards with Fe3+/∑Fe ratios independently characterized. Without using standards to correct spinel Fe3+/∑Fe ratios, uncertainties in log fO2 are roughly double those of samples calculated from corrected spinel Fe3+/∑Fe ratios (Davis et al., 2017, see Methods Appendix and Table 3.1). Despite, or because of, this limitation, we see that ocean island xenoliths record a wide range of fO2, from QFM –2 to nearly QFM +4 with a mean equal to QFM +0.82 (±1.40). This is a much broader range, extending to lower fO2, than compiled by Ballhaus (1993); the mean is within error of that compiled by Mallmann and O’Neill (2007), though again the variance is greater in the present compilation. Of the 143 xenoliths compiled here, 11 were identified by the original authors as pyroxenites (Sen, 1987; Tracy, 1980) which record a more oxidized mean fO2 of QFM +1.44 (±0.63) than the whole set of OIB xenoliths (see red histogram overlain on Fig. 3.3d). The range of fO2 recorded by OIB lavas falls within the range recorded by ocean island xenoliths; however, we cannot draw a genetic relationship between all lithospheric mantle xenoliths and the mantle melts that exhume them at plumes (compare Fig. 3.2e to 3.3d). Presumably, many of these xenoliths represent lithospheric mantle that has experienced limited chemical interaction with plume‐derived melts. Others are likely products of melt‐rock reaction between lithospheric peridotite and plume‐generated melts. The pyroxenite xenoliths were especially likely to have been derived in this way (e.g., Sen & Leeman, 1991), but it is unclear how many of the peridotites were also influenced by plume‐sourced melts. There is another caveat; we do not know the equilibration pressure at the closure temperature of these plume xenoliths. At ridges, we can reasonably infer pressure from peridotite thermometry because geothermal gradients are reasonably well‐characterized. That is not the case in plume settings, where thermal gradients are radial as well as vertical (e.g., Farnetani & Hofmann, 2010). In Figure 3.3e we demonstrate how equilibration at 2.5 GPa, instead of 0.6 GPa, would shift fO2 down by two thirds of a log unit. This illustrates our community’s need for better peridotite mineral barometry.
We venture that a more appropriate comparison might be drawn between peridotitic ocean island xenoliths and ridge peridotites. Both may initiate as residues of melting at ridges, with the former transiting and cooling prior to interacting with a mantle plume. The range of fO2s recorded by ocean island xenoliths encompasses the range of ridge peridotites but skews to higher fO2s by 0.37 log units if we hold pressure constant at 0.6 GPa (tstat = 2.6, tcrit = 2.0, df = 211, p‐value = 0.01). Notably, spinel‐olivine Fe‐Mg exchange (Li et al., 1995) records higher temperatures in the ocean island xenoliths than in the ridge peridotites. This difference in temperature could be the result of heating of the oceanic lithosphere beneath oceanic islands by the plume (Ballhaus, 1993), or it could result from exhumation of these xenoliths from greater depths than the depth of last equilibration experienced by ridge peridotites. In the former case, temperature‐dependent exchange reactions suggest that a mantle parcel preserving a record of hotter conditions should record lower fO2 than a parcel at the same pressure that records cooler conditions (Birner et al., 2018). Consideration of these subsolidus reactions would thus predict ocean island xenoliths to be more reduced than ridge peridotites, in contrast to what we observe. If this interpretation is correct, then the difference in fO2 between ridge peridotites and OIB xenolith source mantle prior to plume heating is even greater than the 0.6 GPa plots in Figure 3.3 suggest, perhaps driven by the interaction of some of these xenoliths with oxidized plume melts. In the latter case, changes in fO2 due to changes in pressure would additionally have to be accounted for to make a direct comparison between ridge peridotites and OIB xenoliths. If no changes in mineral composition or mode are considered, the lower average fO2 calculated assuming a higher pressure of equilibration (Fig. 3.3) suggests that if OIB xenoliths do generally sample deeper portions of the lithosphere than ridge peridotite, then average fO2 of the two are comparable. While exchange reactions and modal changes during ascension and cooling may complicate this relationship, at this time, the data do not suggest that xenoliths recovered from plumes significantly differ in their fO2 compared to peridotites recovered in the ridge setting. Constraints on pressure and effects of temperature on the fO2 recorded by peridotites below their solidus remain poorly understood, and further work is needed to clarify the fO2 signature of xenoliths entrained within plume lavas.
3.3.2. V/Yb Concentrations
V/Yb ratios range from 17–195 in MORB, 60–238 for BABB, and 65–422 in arcs (excluding one arc basalt with a ratio of > 800). We find average V/Yb concentrations in each tectonic setting that are all statistically distinct (p‐values <<< 0.001) with V/Yb of MORB (93 ±17) < BABB (107 ±25) < arcs (158 ±60). We illustrate this in Figure 3.4 with a plot of V/Yb ratios against MgO concentrations in ridge, back‐arc, and arc settings. Translating trace element ratios measured in glass (magmatic liquid) to the oxygen fugacity of the source rock (residue) depends on having accurate, composition‐dependent, mineral‐melt partition coefficients, an accurate knowledge of the source composition, and an accurate melting model (e.g., Canil, 1997; Lee et al., 2003; Lee et al., 2005; Mallmann & O’Neill, 2009; Mallmann et al., 2019). Mallmann and O’Neill (2013), Nicklas et al. (2019), Mallmann et al. (2019), Bucholz and Kelemen (2019), and others have discussed the difficulty in translating V/Sc ratios or olivine‐melt partition coefficients into source fO2s. For example, the fO2 of modern MORB, or Archean mantle, based on V partitioning is up to a log unit higher than that implied by modern MORB Fe3+/∑Fe ratios (Mallmann and O’Neill, 2013; Nicklas et al., 2019). For these reasons, we do not calculate fO2 for each tectonic setting based on the V/Yb ratio, but simply infer the relative oxygen fugacity of each tectonic setting based on the premise that basalt V/Yb will rise with the fO2 of the mantle source that generated the basalt. Under these assumptions, it is clear that fO2 ridges < fO2 back arcs < fO2 arcs.
Figure 3.4 V/Yb ratios of ridge (gray “+” symbols, Gale et al., 2013), back‐arc (blue triangles, Gale et al., 2013), and arc lavas (teal circles, Turner & Langmuir, 2015) as a function of weight percent MgO. We filtered each published data compilation for 6 wt.% < MgO < 12 wt.% and for Dy/Yb < 2 (Laubier et al., 2014). See text for details.
3.4. DISCUSSION
Our compilation and reprocessing of analytical data from the literature yields a synoptic picture of the fO2s recorded by volcanic and mantle rocks across tectonic settings. Some three decades have passed since the seminal compilations from last century (e.g., Ballhaus et al., 1991; Carmichael, 1991; Christie et al., 1986; Frost & Lindsley, 1992; Wood et al., 1990). The volume of data and geographic coverage have increased tremendously and the analytical techniques and activity models have evolved; however, a key finding from those studies remains robust today. Volcanic and mantle rocks from arc settings record significantly higher fO2 than those from ridges. Recycling of oceanic crust and lithosphere back into the Earth at subduction zones generates arc volcanics and related mantle lithologies
