The Osmium Isotope Signature of Phanerozoic Large Igneous Provinces

The Osmium Isotope Signature of Phanerozoic Large Igneous Provinces Book Section How to cite: Dickson, Alexander J.; Cohen, Anthony and Davies, Marc (2021). The Osmium Isotope Signature of Phanerozoic Large Igneous Provinces. In: Ernst, Richard E.; Dickson, Alexander J. and Bekker, Andrey eds. Large Igneous Provinces: A Driver of Global Environment and Biotic Changes. Geophysical Monograph Series. Wiley, pp. 229–246.


INTRODUCTION
The emplacement of Large Igneous Provinces (LIP) is characterized by anomalously high magmatic fluxes, such that the majority of their volume is emplaced within a relatively short time-period of <1-2 million years. They comprise massive volumes of mantle-derived igneous material sometimes in excess of 10 6 km 3 , intruded into the crust as dykes, sills, and batholiths, and extruded onto the surface as effusive lava flows or as explosive ejecta together with a cocktail of superheated gases and fluids in either a subaerial or submarine environment (Coffin & Eldholm, 1992Ernst, 2014). LIPs have often been linked to mantle plumes, which are persistent upwellings of anomalously hot mantle. The rapid eruption rates and huge volumes of material sourced from melting in the plume head during the early stages of LIP emplacement far exceed present-day eruption rates and volumes, and because of this rapidity they are proposed to have had a deleterious impact on the global environment. LIPs occurred periodically throughout Earth's history, at approximate intervals of a few tens of millions of years (Prokoph et al., 2013) and have been associated with episodes of extreme global climate change and biotic extinction (Ernst & Youbi, 2017; Bond & Sun, Chapter 3 this volume; Ernst et al., Chapter 1 this volume) ( Fig. 10.1).
Several key questions regarding LIP behavior have been postulated: When did individual LIP events occur? Was volcanic activity continuous or intermittent over the period of emplacement? Did LIP emplacement drive changes in the global climate system? What climatic feed back processes do LIPs perturb? Do all LIPs alter Earth system processes (weathering, ocean chemistry, warming, etc.) in the same way? Do all LIPs cause biotic extinc tions? The approach to answering many of these ques tions is commonly to undertake rigorous radiometric dating of igneous rocks that can be related to specific epi sodes of LIP activity. These efforts initially centered on Ar-Ar dating of flood basalts. These Ar-dating approaches have been instrumental in establishing firstorder relationships between the timing of LIP events and major extinction and environmental changes across the Phanerozoic and beyond (Wignall et al., 2001), but are sometimes limited by relatively large absolute age uncer tainties, which may be on the order of ~ ±10 5 -10 6 years. Recent advances in U-Pb dating, with reported age uncer tainties of ±10 4 years, have significantly reduced the age uncertainties on some LIP events, thereby allowing a greater understanding of how individual episodes of LIP activity may proceed within a longer period of emplace ment (e.g., Schoene et al., 2010;Svensen et al., 2010;Blackburn et al., 2013;Davies et al., 2017; other contributions in this volume). Ultrahigh-resolution dating of magmatic episodes has greatly improved our understanding of the nature of LIP activ ity, providing a precise framework to explore the poten tial of volcanic activity as a trigger for the complex environmental changes and feedbacks leading to mass extinction events. Nonetheless, this approach requires the preservation of datable rocks that can be stratigraphically related to the overall LIP sequence. This requirement inevitably leads to LIP chronologies that can potentially be discontinuous and patchy.
Despite these limitations, recent advances in radiomet ric dating have shown that the catastrophic environmen tal changes potentially driven by LIP volcanism are triggered by intense short-lived volcanic episodes rather than persistent volcanism spanning the entire period of emplacement. Similarly, the associated feedback pro cesses set in motion (including carbon cycle reorganiza tion, climatic warming, weathering, ocean anoxia, and biotic extinction) operate on comparable centennial-mil lennial timescales. Therefore, in order to constrain the magnitude and duration of these perturbations and establish an order of events, it is necessary to generate proxy data with age constraints precise enough to resolve the environmental changes in the stratigraphic record. Strontium isotopes have been used to constrain the source and duration of weathering during extended warming events, but the long oceanic residence time of Sr (>4 mil lion years; Veizer, 1989) limits its ability to resolve very short-term weathering events. The very much shorter ocean residence time of Os (~10-50 Kyr; Sharma et al., 1997;Levasseur et al., 1999) Barry et al. (2013), Schoene et al. (2015), Duncan et al. (1997), Hofmann et al. (1997), Duncan (2002), Storey et al. (2007), Timm et al. (2011), Blackburn et al. (2013), Loewen et al. (2013), , Davies et al. (2017), and Kingsbury et al. (2018). Dashed lines indicate approximate periods of reduced activity within the overall duration of a LIP.
continually adjusts to the input of Os weathered from newly emplaced volcanic rocks. This feature gives Os-iso tope stratigraphy the almost unique quality of being able to trace the temporal progression of LIP events at fine levels of detail (potentially <10 4 years) at far-field sites that are not affected by the erosion or thermal alteration processes that can disturb sedimentary successions in more proximal settings. It is this utility of Os isotopes that will form the focus of this contribution.

OS ISOTOPE STRATIGRAPHY
Rhenium and Os readily partition into metal, sulfide, and organic phases, and because of this behavior the Re-Os isotope system provides a complementary record of geological processes compared with silicate-hosted iso topic systems such as Rb-Sr, Sm-Nd, Lu-Hf, and U-Pb. Rhenium and Os show differences in compatibility, which give rise to contrasting low and high Re/Os ratios for the mantle and crust, respectively. This marked parentdaughter fractionation and the subsequent radiogenic ingrowth of 187 Os as a result of β-decay of 187 Re produces orders of magnitude variations in the 187 Os/ 188 Os of geo logical reservoirs. In crustal rocks, where the Re/Os ratio is relatively high, the in-situ production of 187 Os leads to high (radiogenic) 187 Os/ 188 Os ratios that average ~1.4 (Peucker-Ehrenbrink & Jahn, 2001). In mantle and ultra mafic rocks, where Re/Os ratios are low, 187 Os/ 188 Os ratios are lower (unradiogenic) with a value that is more chon dritic in nature, ~0.12 (Luck & Allègre, 1983). The oceans record the proportional mixing of the two Os isotope end-members (Peucker-Ehrenbrink & Ravizza, 2000) ( Fig. 10.2).
Three important aspects of the Os-isotope system in regard to LIPs need to be considered. First, Os isotopes are an indirect tracer for LIP activity. The rapid emplacement of LIPs can result in the intense and rapid weathering of juvenile mafic and ultramafic rocks, either by atmospheric and biogeochemical processes, low-tem perature submarine basalt-seawater interaction, or crea tion of hydrothermal systems around submarine volcanic centers (Peucker-Ehrenbrink & Ravizza, 2000;Cohen & Coe, 2002;Turgeon & Creaser, 2008 Seawater Levasseur et al. (1998) Sharma et al. (1997 River water Levasseur et al. (1999) Pegram et al. (1994 Upper continental crust (estimated) Ravizza and Turekian (1992) Burton et al. (1999) Meteorites/mantle Luck et al. (1980) Luck and Allègre (1983) Walker and Morgan (1989) 187 Os/ 188 Os 0 0 .5 1 1 .5 2 2.5 Figure 10.2 The modern mass balance of osmium. Osmium isotope ratios for different geological materials are shown along with key literature references. The modern seawater value of ~1.06 represents a contribution of ~10% to 30% from unradiogenic (mantle and extraterrestrial dust) fluxes and ~70% to 90% from radiogenic sources in the continental crust.  Klemm et al. (2008). The grey-shaded region denotes a shift toward more unradiogenic seawater 187 Os/ 188 Os ratios that are taken to record the rapid weathering of basalts associated with LIP emplacement and/or a reduction in the weathering rate of continental rocks.

Recent ocean sediments
"signature" of a LIP might involve swings in seawater 187 Os/ 188 Os in either direction at different times, a feature that is clearly apparent during many LIP events, for example, the Ontong-Java Plateau and the North Atlantic Igneous Province (Bottini et al., 2012;Dickson et al., 2015). The second important feature of the Os isotope system is the short residence time of Os in the oceans, of ~10-50 kyrs (Sharma et al., 1997;Levasseur et al., 1999). This feature provides Os-isotope stratigraphy with the potential for tracing the pulsed emplacement of LIPs at timescales of 10 3 -10 4 years, and the nature of the associ ated climate/weathering feedbacks. As will be discussed, this utility has rarely been fully exploited for any indi vidual LIP to date. Finally, the Os-isotope composition of the oceans can also be influenced by extraterrestrial fluxes, such as from impactor events (e.g., Sato et al., 2013) or cosmogenic particles (Ravizza, 2007). Extraterrestrial impacts have been suggested for several intervals bracketed by LIPs, and these must be borne in mind when examining the reconstructed temporal evolution of seawater chemistry.
Osmium is present in seawater only in ultra-trace con centrations but is strongly enriched in reducing marine sediments. The enrichment of Os (and Re) in low-oxygen depositional settings means that paleo-seawater 187 Os/ 188 Os ratios can be traced by the careful measure ment of Re and Os compositions in organic-rich mudrocks, followed by a correction for the postdeposi tional decay of 187 Re (Ravizza & Turekian, 1989Cohen et al., 1999). The requirement to correct 187 Os/ 188 Os ratios in organic-rich rocks for 187 Re decay means that socalled initial Os-isotope stratigraphies (Os i ) have been produced mainly for those events where suitable deposits exist with independent age control. Where there is no independent age control, "initial" 187 Os/ 188 Os ratios can be estimated using the isochron approach of measuring samples with different 187 Re/ 188 Os ratios collected from a restricted stratigraphic range (e.g., Cohen et al., 1999). Some Os-isotope records have also been produced from Fe-Mn crusts and oxic metalliferous sediments (e.g., Pegram et al., 1992;Peucker-Ehrenbrink et al., 1995;Burton et al., 1999;Ravizza et al., 2001;Klemm et al., 2005Klemm et al., , 2008Burton, 2006;Robinson et al., 2009). These records do not require a significant correction for Re decay, but may be systematically biased due to the partial liberation of detritally hosted Os phases from the bulk sediment (e.g., Pegram & Turekian, 1999). Furthermore, such records from Fe-Mn crusts have a limited temporal resolution because of their slow accumulation rates. In all types of approach (mudrocks or crusts/sediments), Os i records spanning early Phanerozoic LIPs (i.e., pre-Permian) have not yet been widely produced. The LIP record of only the late Phanerozoic will therefore be sum marized in the following discussion and illustrated in

The Columbia River LIP (~17-15 Ma)
As the youngest LIP of the Phanerozoic, the Columbia River (CR) event has a very well defined chronology and stratigraphic framework (Barry et al., 2013;Riedel et al., 2013). 40 Ar/ 39 Ar and K-Ar age determinations of the CR eruptive history suggested that activity occurred across a total interval of ~16.9-6 Ma, with most activity occur ring during emplacement of the Grande Ronde basalt, from ~16-15.6 Ma ( Barry et al., 2013). Recent zircon U-Pb ages of CR ashes have refined this chronology to con strain ~95% of the eruptive history to the interval 16.7-15.9 Ma (Kasbohn & Schoene, 2018). There are few Os-isotope data that record the impact of the CR LIP on ocean chemistry. The records that exist are from oceanic ferromanganese crusts, which record changes in seawater 187 Os/ 188 Os at multimillion-year timescales that are far in excess of the oceanic residence time of Os (Klemm et al., 2005(Klemm et al., , 2008Burton, 2006). These records (illustrated in Fig. 10.3) do suggest a small 187 Os/ 188 Os shift of ~0.1 toward more unradiogenic ratios in seawater during the Miocene, as would be expected from an enhanced weath ering flux of unradiogenic Os from CR basalts (Klemm et al., 2008). However, the timing of this shift depends on the age-model applied to the crust records. Even revised age-models based on Os-isotope stratigraphy imply an unradiogenic shift in 187 Os/ 188 Os between ~15 and 12 Ma, a pattern that significantly postdates radiometric ages of most of the CR eruptive episodes. For a LIP to have had a discernable impact on Os ocean chemistry, it must have been volumetrically large, the constituent lavas and intru sive rocks must have contained high Os concentrations, and the rocks must have been weathered rapidly following emplacement. Although effusion rates in individual pulses of CR volcanism may have been comparable to larger LIPs in the geological record, the CR river event was volumetrically small compared with many earlier Phanerozoic LIPs, and the amount of basalt weathered was orders of magnitude smaller than, for example, the CAMP event at the Triassic-Jurassic boundary (Cohen & Coe, 2002). Thus, it is possible that the putative unradio genic signal observed by Klemm et al. (2008) is actually unrelated to the CR LIP, and records a different pertur bation to Os ocean chemistry in the Miocene.

Ethiopian-Yemeni Flood Basalts (~31-29 Ma)
The Ethiopian-Yemeni LIP has the best-preserved sequence of flood basalts in the Cenozoic geological record. The main phase of flood basalt volcanism began shortly before ~30 Ma and lasted for less than ~1 million years (Hofmann et al., 1997;Ukstins et al., 2002) before continuing in pulses concomitant with the opening of the Red Sea and the Gulf of Aden (Courtillot & Renne, 2003). Magnetostratigraphy of the Ethiopian flood basalts indicates a correlation to magnetochrons C11r to C11n (Hofmann et al., 1997;Touchard et al., 2003), mak ing them younger than the Eocene-Oligocene boundary event, which occurred during chrons C13r-C12r (~34 Ma; e.g., Zachos et al., 1996). The effect of the Ethiopian-Yemeni LIP on the osmium chemistry of the oceans is not well understood, with low-resolution data available from only three locations (Peucker-Ehrenbrink & Ravizza, 2012). These records agree in the sense that they all show 187 Os/ 188 Os ratios evolving to less radiogenic val ues at ~30-31 Ma. However, the magnitude and pattern of this decrease varies, from ~0.08 in Indian Ocean ODP Site 711, to ~0.12 in South Atlantic DSDP Site 522 (Peucker Ehrenbrink & Ravizza, 2012). Os-isotope stra tigraphy therefore appears to reveal a signature of basalt weathering on ocean chemistry, though the true size of this weathering flux and its wider temporal context are limited by the available data.

The North Atlantic Igneous Province (NAIP) (~61-54 Ma)
The emplacement of the NAIP near the Paleocene-Eocene boundary has been suggested to have influenced the genesis of rapid global warming during the Paleocene-Eocene Thermal Maximum (PETM; Storey et al., 2007;Frieling et al., 2016), an event that also includes an extinc tion of benthic foraminifera (Thomas & Shackleton, 1996). The NAIP consists of a series of subaerial lava flows and intrusive units (e.g., Svensen et al., 2004Svensen et al., , 2010 that are dated to between ~60 and ~53 Ma (Storey et al., 2007;Svensen et al., 2010;Wilkinson et al., 2017). Only two existing Os-isotope records (from Fe-Mn crusts CD-29 and D11-1) cover the entire period of emplace ment (Klemm et al., 2005;Burton, 2006). These records show a shift in seawater 187 Os/ 188 Os to slightly more unra diogenic values, as would be expected as extruded basalts began to weather into the oceans (Fig. 10.4). A number of high-resolution Os-isotope records span the Paleocene-Eocene boundary (Ravizza et al., 2001;Weiczorek et al., 2013;Dickson et al., 2015), when the accumulation rate of NAIP basalts increased significantly at the commence ment of seafloor spreading (Storey et al., 2007b). These records actually show a small change (~0.05) to more radiogenic 187 Os/ 188 Os ratios that has been interpreted to reflect enhanced weathering of terrestrial rocks due to elevated atmospheric temperatures and moisture (Ravizza et al., 2001;Dickson et al. 2015). The small magnitude of the increase in 187 Os/ 188 Os compared with other Phanerozoic events (e.g., Cohen et al., 2004) may be due to the competing influences of Os being weathered from both radiogenic and unradiogenic sources at the same time. Several of the NAIP data sets also demonstrate a very brief shift to more unradiogenic values in seawater near the Paleocene-Eocene boundary, which likely records a pulse of unradiogenic Os associated with magmatic activity at the commencement of the PETM (Weiczorek et al., 2013;Dickson et al., 2015), or perhaps an extrater restrial impact event (c.f. Schaller et al., 2017). The strati graphic correspondence between unradiogenic Os-isotope ratios and a peak in Hg concentrations in pre-PETM deposits in Svalbard tend to support a volcanic origin for this feature (Jones et al., 2019). This observation high lights the potential for Os isotope stratigraphy to reveal very fine-scale detail of volcanic activity that is pertinent to testing hypotheses relating LIP emplacement to rapid climate change.
The influence of episodic NAIP activity on brief global warming events (hyperthermals) that occurred after the PETM is largely untested. A new Os-isotope record is shown in Figure 10.4 (data in Table 10.1) from IODP Site M0004A (Arctic Ocean) spanning one such event, Eocene Thermal Maximum 2 (~53 Ma). These data were produced using techniques identical to those of Dickson et al. (2015). Initial Os-isotope ratios increase by ~0.1 (0.38-0.48) shortly before the carbon isotope excursion that marks the start of the event, and again in more dramatic fashion at the termination of the carbon cycle perturbation, from ~0.4 to 0.8. The data are similar to 187 Os/ 188 Os ratios of metalliferous sediments from DSDP 549 that contain a shift to more radiogenic values (from ~0.44 to 0.50) across ETM 2 (Peucker-Ehrenbrink & Ravizza, 2012), thus sup porting the hypothesis of a rapid increase in continental weathering across the hyperthermal. A single unradiogenic value of 0.18 also stratigraphically precedes ETM 2 at Site M0004A ( Fig. 10.4). Given the short duration of the ETM 2 (~100 kyrs) the unradiogenic value before the event began is at least consistent with a volcanic trigger. These observa tions come with the caveat of increasing hydrographic restriction in the Arctic during the Early Eocene (Brinkhuis et al., 2006;Dickson et al., 2015) that may have caused the 187 Os/ 188 Os ratio of Arctic Ocean seawater to deviate from the global value, although the comparison of ETM 2 data from Site M0004A and Site 549 suggest that this effect was small. The NAIP is a clear candidate for future high-reso lution Os-isotope studies that seek to unravel the interac tion of volcanism and climate change in the early Cenozoic.

The Deccan Traps (~66.3-65.5 Ma)
Ar-Ar age estimates of the timing of LIP emplacement place the Deccan Traps close to the mass-extinction event at the Cretaceous-Paleogene boundary (K-Pg)   2015); ETM-2 Os isotope data are presented in this paper; ETM-2 C-isotope data for IODP Site M0004A are compiled from Sluijs et al. (2009) and Dickson and Cohen (2012). Grey-shaded regions denote the main phases of the PETM and ETM-2. (Chenet et al., 2007). U-Pb dating of zircons recovered from ashfall and erosive units interbedded with lava flows has recently allowed the chronology of this event to be improved substantially, with an estimate of ~753 kyrs for 80%-90% of the total eruptive history (Schoene et al., 2015). Os-isotope records from a variety of carbonate successions show reproducible trends in seawater 187 Os/ 188 Os prior to the K-Pg, with a ~20%-25% decrease toward unradiogenic values commencing at the C29r/ C30n boundary, followed by a second decrease toward even more unradiogenic values occurring much closer to the K-Pg itself (Fig. 10.5; Ravizza & Peucker-Ehrenbrink, 2003;Robinson et al., 2009). Os data from K-Pg loca tions in Europe and the United States show that seawater 187 Os/ 188 Os decreased nearly to mantle values of ~0.14 at the acme of the unradiogenic shift prior to the K-Pg. This decrease is consistent both with an extraterrestrial impac tor (Luck & Allègre, 1983;Esser & Turekian, 1989;Geissbühler, 1990;Peucker-Ehrenbrink et al., 1995;Meisel et al., 1995;Yin et al., 1995), and also with the emplacement of the Poldapur Deccan basalts, according to recent U-Pb data (Schoene et al., 2019). Osmium isotope data from latest Cretaceous rocks have been instrumental in testing the hypothesis that the eruption of the Deccan Traps caused the end-Cretaceous mass extinction. The earliest shift in 187 Os/ 188 Os associated with volcanism occurs considerably earlier than the major extinction horizon, an observation that tends to favor an extraterrestrial impactor as the cause of most (though not all) of the major biotic consequences of this time interval (Ravizza & Peucker-Ehrenbrink, 2003).
The Deccan Traps illustrate an interesting conundrum in the interpretation of 187 Os/ 188 Os data in terms of LIP activity. Ravizza and Peucker-Ehrenbrink (2003) inter preted the decrease in 187 Os/ 188 Os commencing at the C29r/C30n boundary as recording a decrease in the weathering of radiogenic Os as flood basalts were extruded across crystalline basement rocks in the early phases of the Deccan LIP. This argument was in part supported by the low concentration of Os in Deccan basalts . Other LIP events featuring a decrease in the 187 Os/ 188 Os of seawater have been inter preted in terms of an increase in the weathering of unra diogenic Os from mafic rocks (Turgeon & Creaser, 2008;Tejada et al., 2009;Bottini et al., 2012;Du Vivier et al., 2014). These differences highlight the fact that the marine 187 Os/ 188 Os ratio is the product of two competing inputs, and that the use of complementary data sets is often required to arrive at a satisfactory interpretation of the observed chemostratigraphic variations.

The Caribbean (~95-60 Ma), High Arctic (127-81 Ma), and Madagascan (~84-95 Ma) LIPs
Volcanism has long been hypothesized as a trigger for one of the most profound episodes of Phanerozoic ocean deoxygenation, at the Cenomanian-Turonian boundary. This event, Oceanic Anoxic Event 2 (OAE 2), took place at a similar time as the emplacement of several LIPs, most notably the Caribbean LIP (CLIP) and the High Arctic LIP (HALIP). Early studies attributed concentra tion spikes of mafic-derived trace elements in sedimen tary rocks to infer volcanism (e.g., Orth et al., 1993), and this approach has continued recently (Eldrett et al., 2014). The publication of the first Os-isotope records for the Cenomanian-Turonian boundary, by Turgeon and Creaser (2008), revealed mantle-like signatures in global seawater that were sustained for hundreds of thousands of years during the acme of the environmental changes associated with OAE 2. These data firmly supported the significant role of voluminous volcanic activity in driving and sustaining widespread environmental change during this event, presumably through volcanism-climate feed backs. Such feedbacks may have included the delivery of biolimiting nutrients and sulfate to the oceans, stimulat ing organic matter production and the consequent con sumption of dissolved oxygen in many parts of the oceans (Adams et al., 2010;Jenkyns, 2010). The Os-isotope data sets of Turgeon and Creaser (2008) have since been sup plemented by Du Vivier et al. (2014Vivier et al. ( , 2015, who were able to show how abrupt shifts in 187 Os/ 188 Os (i) ratios toward  Robinson et al. (2009). The greyshaded region denotes a shift toward more unradiogenic seawater 187 Os/ 188 Os ratios that are taken to record the rapid weathering of basalts associated with LIP emplacement and/or a reduction in the weathering rate of continental rocks. mantle values (~0.15) occurred several thousands of years in advance of the positive carbon isotope excursion that defines the event (c.f. Tsikos et al., 2004) (Fig. 10.6). The persistence of unradiogenic Os-isotope values over a period of several hundreds of thousands of years, far in excess of the Os residence time in the oceans, demon strates the prolonged period of time during which Os was weathered from volcanic rocks on land and/or by subma rine basalt-seawater interaction (Turgeon & Creaser, 2008;Du Vivier et al., 2014. Furthermore, changes in the concentration of Os in sedimentary successions throughout the phase of otherwise unradiogenic Os-iso tope ratios within OAE 2 may suggest small changes in the amount of Os being weathering into the oceans (Du Vivier et al., 2014. However, despite the clear sig nature of LIP activity afforded by the unradiogenic Osisotope values that span OAE 2, these data are not able to unambiguously fingerprint the source of the unradio genic Os flux. Various studies continue to debate the rela tive importance of volcanism associated with the HALIP (Eldrett et al., 2017) and the CLIP (Kuroda et al., 2007;Holmden et al., 2016;Scaife et al., 2017), while Ar-Ar ages for eruptive events associated with the Madagascan LIP (Cucciniello et al., 2010) also slightly overlap the age of the Cenomanian-Turonian boundary (Meyers et al., 2012). It is possible that these events all contributed in some way to the widespread environmental changes that occurred during OAE 2.
OAE 2 provides an interesting case study of hysteresis in Earth-system processes. The lead-lag relationship between Os-isotope and C-isotope changes at the onset of OAE 2 clearly supports the contention of a volcanic trigger with the rapid emplacement of submarine basalts, probably associated with the CLIP, being rapidly weath ered into seawater. However, the shift to more radiogenic Os-isotope ratios in marine sediments before the end of OAE 2 does not clearly link to a decrease in global tem peratures, as would be expected if volcanic CO 2 emissions slowed and were further reduced by silicate weathering and organic-carbon burial feedbacks (Robinson et al., 2019). As well as driving transient environmental changes, LIP volcanism may also drive Earth's climate system into new, quasi-stable states.

The Ontong-Java Plateau (~126-117 Ma)
Os-isotope stratigraphy has been instrumental in dem onstrating that the emplacement of the Ontong-Java LIP in the ancestral Pacific Ocean occurred at precisely the same time as an episode of major environmental change, during Oceanic Anoxic Event 1a (Tejada et al., 2009;Bottini et al., 2012). These Os-isotope records, from loca tions in two different ocean basins, have strikingly similar 187 Os/ 188 Os (i) values, and exhibit near-identical shifts in 187 Os/ 188 Os (i) ratios when compared with the carbon-and bio-stratigraphic frameworks for each locality (Malinverno et al., 2010) (example in Fig. 10.7). The Osisotope records clearly support three major findings. First, the major phase of environmental change during OAE 1a (the "Selli" level; c.f. Coccioni et al., 1987) coin cided with almost mantle-like 187 Os/ 188 Os (i) ratios of ~0.15-0.2 (Bottini et al., 2012). These unradiogenic val ues must have been maintained for almost 900,000 years by the continual hydrothermal weathering of very large quantities of mafic and ultramafic rocks during a major phase of submarine LIP emplacement. Second, the 187 Os/ 188 Os (i) records bear some similarity to events sur rounding Late Cretaceous OAE 2 because the influence of LIP weathering on ocean chemistry began prior to the onset of OAE 1a (Bottini et al., 2012). This lead-lag rela tionship implies a causal relationship between the O-J LIP and major environmental change during OAE 1a. Somewhat enigmatic, less radiogenic 187 Os/ 188 Os (i) ratios in Upper Barremian strata of the Cismon core, Italy, hint at an even earlier, less-intense phase of volcanism that may have been linked to a biocalcification crisis in nannofossil flora Bottini et al., 2012). The existing data resolution is, however, not sufficient to resolve this hypothesis. Third, 187 Os/ 188 Os (i) ratios exhibit significant CENOMANIAN δ 13 C org Figure 10.6 Cenomanian-Turonian Os-isotope data spanning Oceanic Anoxic Event 2, and a putative magmatic episode linked to the Caribbean LIP and/or the High Arctic LIP. C-isotope data are from Erbacher et al. (2005) and Os-isotope data are from Turgeon and Creaser (2008). The grey-shaded region denotes a shift toward more unradiogenic seawater 187 Os/ 188 Os ratios that are taken to record the rapid weathering of basalts associated with LIP emplacement and/or a reduction in the weathering rate of continental rocks.
shifts throughout OAE 1a, with radiogenic values occur ring at the base of the Selli level at Gorgo a Cerbara (Italy; Tejada et al., 2009) and Cismon (Bottini et al., 2012). As with all 187 Os/ 188 Os (i) records, such shifts cannot be uniquely interpreted as a reflection of continental weathering because of the interplay between the weather ing of both mafic and continental rocks. Circumstantial reasoning based on coeval proxy data, however, can assist with qualitative interpretations of such stratigraphic fluc tuations in 187 Os/ 188 Os (i) .

The Karoo-Ferrar LIP (189-178 Ma)
Volcanism in the Karoo and Ferrar (K-F) provinces, in present-day South Africa, South America, and Antarctica, has been linked to episodes of rapid environ mental change in the Early Toarcian, particularly the Toarcian Oceanic Anoxic Event (T-OAE). Radiometric dating of rocks attributable to the Karoo and Ferrar provinces indicate an overall duration of several millions of years (e.g., Duncan et al., 1997;Sell et al., 2014), with a much shorter period of intense activity around the T-OAE itself, including hydrothermal venting, at ~183 Ma (e.g., Svensen et al., 2007Svensen et al., , 2012. Three marine Os (i) isotope records span part of the dura tion of the K-F LIP, from Yorkshire, UK (Cohen et al., 2004); North Wales, UK (Percival et al., 2016); and western Canada (Them et al., 2017). A fourth Os-isotope data set was recently measured in lacustrine shales, of the Chinese D'hanzhai member, which contain distinctly more radiogenic values than the marine sections (Xu et al., 2017). The marine records all retain a similar range of values, which point to a well-mixed ocean inventory of Os in the early Jurassic ( Fig. 10.8). They also all exhibit large increases in 187 Os/ 188 Os (i) ratios toward radiogenic values across the T-OAE in the range 0.3 to 0.8 (Fig. 10.8), which is somewhat counterintuitive given the intense vol canic activity that took place at that time. The explana tion for this trend may lie partly in the largely subaerial nature of the K-F LIP and the associated lack of conti nental breakup, which inhibited the weathering rate of mafic rocks into the oceans following emplacement (Percival et al., 2016). Also, volcanic-driven climate warming probably contributed to the intense weathering of continental crust that would have caused a large influx of radiogenic Os to the oceans, thereby overwhelming any unradiogenic flux (Cohen et al., 2004). The Os iso tope records spanning part of the K-F LIP clearly dem onstrate the unique characteristics of the climatic and weathering feedbacks associated with this LIP, certainly in contrast with large LIP events of the Cretaceous (Ontong-Java and Caribbean LIPs) (Figs. 10.6 and 10.7). However, no single Os isotope record has yet been pro duced that spans the entire estimated K-F duration. Establishing such records in the future will be useful to test hypotheses linking volcanism to early Jurassic cli mate change, particularly in light of recent studies that tend to suggest ocean deoxygenation began considerably earlier than the T-OAE, closer to the putative onset of K-F volcanism before ~183 Ma (Them et al., 2018).

The Central Atlantic Magmatic Province (201.6-200.9 Ma)
The Central Atlantic Magmatic Province (CAMP) has been the subject of many detailed studies that have estab lished precise chronologies of volcanic pulses from ~201.6 to 200.9 Ma (e.g., Schoene et al., 2010;Blackburn et al., 2013;Davies et al., 2017). These, and other stratigraphic studies, have shown a close temporal relationship between initial pulses of intrusive volcanism, a first-order mass extinction, and a large negative carbon isotope excursion of only a few tens of kyrs duration (Hesselbo et al., 2002;Ruhl et al., 2010;Whiteside et al., 2010). While there are two Os (i) isotope records that span the duration of the CAMP, from the southwest UK (Cohen & Coe, 2002) and from Japan (Kuroda et al., 2010), neither of these is presently able to resolve the impact of the CAMP on sea water chemistry with comparable temporal precision to U-Pb chronologies (e.g., Blackburn et al., 2013) or chemostratigraphic studies (e.g., Hesselbo et al., 2002; Tejada et al. (2009). The greyshaded region denotes a shift toward more unradiogenic seawater 187 Os/ 188 Os ratios that are taken to record the rapid weathering of basalts associated with LIP emplacement and/or a reduction in the weathering rate of continental rocks. Ruhl et al., 2011). Both Os records contain evidence for a decrease in the 187 Os/ 188 Os ratio of seawater between the latest Triassic (Rhaetian) and the early Jurassic (Hettangian) broadly from ~0.3-0.6 to 0.1-0.5. However, the UK record has no data from the critical interval encompassing the earliest emplacement of intrusive mag mas and the end-Triassic mass extinction, corresponding to the upper Westbury and Lilstock Formations (Cohen & Coe, 2002) (Fig. 10.9). The Japanese record, in con trast, has a higher stratigraphic resolution, but differs from the UK record in two respects: first, minimum 187 Os/ 188 Os (i) ratios of ~0.2 occur in late Triassic deposits in Japan, but in early Jurassic deposits in the UK; and second, 187 Os/ 188 Os (i) ratios are slightly more radiogenic throughout the Japanese section than in the UK. The stratigraphic differences between the sections may be due to the difficulty of correlating the Japanese locality with the northern European biostratigraphic scheme (Kuroda et al., 2010), or to heterogeneity in late Triassic-early Jurassic seawater 187 Os/ 188 Os ratios. As with the K-F LIP, the potential for Os-isotope chemostratigraphy to unravel the impact of the CAMP on global seawater chemistry at a resolution comparable to the U-Pb geochronology has not yet been fully realized.

The Siberian Traps (~252.2-250.2 Ma)
Zircon U-Pb ages of the Siberian Traps have been used to formulate a model of LIP evolution in relation to the largest mass extinction event of the Phanerozoic Eon.
Volcanism began at ~252.4 Ma and ended before ~250.2 ± 0.3 Ma (Kamo et al., 2003;Bowring et al., 1998;. The mass extinction horizon itself has been tied closely to the first  Figure 10.9 Triassic-Jurassic Os-isotope data spanning the emplacement of the Central Atlantic Magmatic Province and the end-Triassic extinction event. Data are from Cohen and Coe (2002). The grey-shaded region denotes the main period of CAMP emplacement. Note that the existing Somerset Osisotope record does not contain data from within this zone.
evidence for intrusive volcanism at ~ 251.9 Ma (Burgess et al., 2017). Despite the detailed chronological con straints on the Siberian Traps, osmium isotope stratigra phy has not yet been applied to this event in great detail. 187 Os/ 188 Os (i) values have been estimated from Re-Os isochrons with ages of ~252-252.5 Ma, to the range of 0.56 to 0.62 (Georgiev et al., 2011) (Fig. 10.10). These data tend to suggest a limited evolution of seawater 187 Os/ 188 Os across the early interval of volcanism, although younger Re-Os data sets (247-234 Ma) have 187 Os/ 188 Os (i) values of ~0.8-1.4, implying a slight evolution of seawa ter to more radiogenic values by the early Triassic (Yang et al., 2004;Xu et al., 2009;Pašava et al., 2009). Nonetheless, the data are of low resolution and do not clearly cover the major phase of intrusive volcanism, allowing for the large uncertainties in the Re-Os estimates of depositional ages and 187 Os/ 188 Os (i) (Georgiev et al., 2011). A single stratigraphic record at Opal Creek, Alberta, exhibits a decrease in 187 Os/ 188 Os (i) from 0.54 to 0.35 following the Late Permian extinction level, followed by a return to ~0.60 at the Permian-Triassic boundary (Schoepfer et al., 2012) (Fig. 10.10). These data suggest a transient excursion in seawater Os chemistry toward unradiogenic 187 Os/ 188 Os values during an interval of intrusive volcanism in the Siberian Traps (c.f. Burgess et al., 2017), which is temporally constrained by U-Pb ages to a duration <60 kyrs (c.f. Burgess et al., 2014).
Additional data for this section show a small excursion to an unradiogenic value of ~0.20 shortly following the main extinction level (Georgiev et al., 2015). The reasons for this brief excursion are not clear but may be linked to variations in subaerial basalt weathering and/or the transport of radiogenic or unradiogenic weathering sig nals from the Siberian LIP to the global ocean. The rela tively small magnitude of documented 187 Os/ 188 Os (i) variation for the Siberian Traps, compared with, for example, the CAMP or OJP LIPs, may be related to the high latitude of their emplacement, which would have resulted in a low weathering rate. Thus, the impact of the Siberian Traps on ocean Os chemistry may have been dis proportionate to their considerable environmental impact.

SUMMARY: OS-ISOTOPE STRATIGRAPHY AND LIPS
Os isotope data from a number of marine sedimentary successions have been instrumental in establishing the link between LIP volcanism and major environmental changes in the Early Cretaceous (OAE 1a) and Late Cretaceous (OAE 2), with significant shifts in seawater chemistry to mantle-like values over timescales of 10 4 -10 5 years (Turgeon & Creaser, 2008;Tejada et al., 2009;Bottini et al., 2012;Du Vivier et al., 2014. The data have been useful in distinguishing the effect of extrater restrial impacts and LIP volcanism on major extinction events (Ravizza & Peucker-Ehrenbrink, 2003;Robinson et al., 2009) and in testing hypotheses of Earth-system feedbacks during LIP-inspired intervals of profound global warming (Cohen et al., 2004;Dickson et al., 2015;Percival et al., 2016;Them et al., 2017).
Nonetheless, the usefulness of Os-isotope stratigraphy to understand the timing and effects of LIP volcanism is capable of further refinement. Os isotopes have most commonly been used to investigate the effects of volcan ism on stratigraphically well-defined episodes of environ mental change, thus testing putative links between volcanism and climate. While these groundbreaking stud ies established the fundamental basis for this approach, the resulting data do not often span the full interval of LIP activity. Additionally, the fundamental drivers of the Os-isotope budget of the oceans (the weathering of con tinental and oceanic rocks) make it difficult to uniquely disentangle the effects of basalt and continental rock weathering on seawater 187 Os/ 188 Os ratios. Many different factors may cause the balance of the weathering products of these sources to vary over time, in addition to the simple presence of a LIP. Such factors include the preva lence of intrusive or extrusive volcanism; the amount and composition of Os liberated from igneous and sedimen tary rocks during intrusive events; the latitude of LIP  Figure 10.10 End-Permian Os-isotope data spanning the emplacement of the Siberian Traps LIP. Circles for the Siberian Traps are for three 187 Os/ 188 Os (i) estimates (Georgiev et al., 2011) whose age uncertainties span the entire duration of LIP emplacement (c.f. Burgess & Bowring, 2014). Additional data (squares) are from Schoepfer et al. (2012). emplacement (and thus proximity to regions of intense physical or chemical weathering); the association of LIP emplacement with continental breakup (i.e., the CAMP); and the effects of volcanism-climate feedbacks on the intensity and/or congruency of global weathering regimes. The heterogeneous Os isotope signatures of Phanerozoic LIPs reflect the many ways in which these factors can combine in both time and space. The recent emergence of mercury (Hg) concentrations as a proxy for volcanism (e.g., Sanei et al., 2012;Percival et al., 2015;Charbonnier et al., 2017;Scaife et al., 2017;Percival et al., Chapter 11 this volume) opens the possi bility of pairing Os-isotopes with Hg analyses to help interpret stratigraphic variations in 187 Os/ 188 Os (i) (e.g., Percival et al., 2016) (Fig. 10.8). Additionally, the role of continental rock weathering as a driver of Os-isotope change in the oceans might also be better constrained by pairing these measurements with other emerging proxies for silicate weathering, such as Li isotopes. One such application of Os and Li isotopes, to OAE 2, has been used to calculate the mass of basalts weathered into the oceans during that event (Pogge van Strandmann et al., 2013). Paired applications of Os and Li may yield fruitful insights into global weathering patterns related to other LIP events.
Os-isotope stratigraphy holds a crucial place in the compendium of approaches used to investigate the effect of LIPs on the Earth's environment. Future applications of this technique at even finer temporal scales will offer further insights into the timing and environmental conse quences of LIPs.