8 Preliminary Appraisal of a Correlation Between Glaciations and Large Igneous Provinces Over the Past 720 Million Years

Earth has gone through periods of cooling including global, near global, or regional glaciations, which are observed in the Archean, Paleoproterozoic, Neoproterozoic, Ordovician, Permo‐Carboniferous, and Cenozoic times. We review the mechanisms by which large igneous provinces (LIPs) and silicic LIPs (SLIPs) can cause global cooling. Then we investi­ gate the correlation of LIPs with important glaciation events, focusing on those of Neoproterozoic and Phanerozoic age. The 720 Ma Franklin‐Irkutsk LIP, a large part of which was emplaced into an evaporite basin and all of which was emplaced in the tropics, is linked with the start of the Sturtian glaciation, one of the longest and most severe glaciations in Earth history. The ca. 579 Ma pulse of the Central Iapetus Magmatic Province (CIMP) is associated with the start and end of the Gaskiers glaciation. The Hirnantian glaciation (ca. 440 Ma) may be associated with poorly dated ca. 450 Ma intraplate magmatism in several regions, including eastern Siberia, South Korea, Argentina, Iran, and elsewhere. It is also coincident with a huge volume of silicic volcanic provinces generated by supereruptions. Permo‐Carboniferous glacia­ tions (P1–P4, 300–260 Ma) can be correlated with widespread intraplate magmatism of the European northwest African magmatic province (and its initiation as the 300 Ma Skagerrak LIP), and also the 259 Ma Emeishan LIP of China. A recently recognized ca. 34 Ma initial pulse of the Afro‐Arabian LIP matches the Eocene‐Oligocene cooling (Oi‐1 glacia­ tion). More precise dating of both the LIPs and cooling events is required to confirm the correlations and to assess the role of LIPs relative to other causes proposed for global and regional glaciations.


LIPs and Their Climatic Effect
Large igneous provinces (LIPs) represent large volume (>0.1 Mkm 3 ; frequently above ~1 Mkm 3 ), mainly mafic (ultramafic) magmatic events of intraplate affinity, which occur in both continental and oceanic settings, and are typically of short duration (<5 Myr) or consist of multi ple short pulses over a maximum of a few tens of Myr (Coffin & Eldholm, 1994, 2005Bryan & Ernst, 2008;Bryan & Ferrari, 2013;Ernst, 2014;Ernst & Youbi, 2017, Ernst et al., Chapter 1 this volume; and references therein). They comprise volcanic packages (flood basalts), and a plumbing system of mafic dyke swarms, sill complexes, mafic-ultramafic layered intrusions, and a lower crustal magmatic underplate. LIPs can also be associated with silicic magmatism (including dominantly silicic events termed silicic LIPs, or SLIPs, sometimes including the socalled supereruptions/supervolcanoes), and also carbon atites and kimberlites.
LIPs and SLIPs can have significant global climatic effects including causing mass extinction events, via a complex web of changes (characteristically rapid) in atmospheric/oceanic acidification, oceanic anoxia, sea level, toxic metal input (e.g., Hg), and, most relevant to this appraisal, both warming and cooling (e.g., Rampino et al., 1988;Stothers, 1993;Courtillot et al., 1996;Wignall, 2001Wignall, , 2005Courtillot & Renne, 2003;Kelley, 2007;Bond & Wignall, 2014;Rampino & Self, 2015;Burgess & Bowring, 2015;Bond & Grasby, 2017;Rampino & Caldeira, 2017). In the broad est sense, LIPs can affect (or even induce) shifts between icehouse, greenhouse and supergreenhouse climatic states (e.g., Kidder & Worsley, 2010, 2012). An important con cept is that while LIP events are large (often millions of square kilometers in horizontal extent), they are not global in extent. But the important insight is that they can have a global effect on the environment both atmos pheric and oceanic (e.g., . We first review the range of mechanisms by which LIPs and SLIPs can cause global cooling, and then consider the evidence for specific global cooling events to be linked with LIPs. We also consider which LIP events could be responsible for the termination of cooling events. Table 8.1 provides a summary of Neoproterozoic and Phanerozoic glaciations and summarizes inferred origins and speculates on potential links with LIPs in each case. In most cases, the links we propose are plausible but not proven. High precision geochronogy (down to uncer tainly levels to <50,000 yr) is required for fully testing the relationships as has been achieved for LIPs such as the 252 Ma Siberian Traps, the 201 Ma Central Atlantic Magmatic Province (CAMP), and the 66 Ma Deccan Traps, which now have compelling links with end Permian, end Triassic and end Cretaceous mass extinc tions, respectively (e.g., Burgess et al., 2015b;Davies et al., 2017;Schoene et al., 2015;Kasbohm et al., Chapter 2 this volume).
In this contribution, we test the hypothesis that glacia tions correlate with LIPs; or stated conversely, we test the null hypothesis that there is no correlation between cli matic cooling and LIPs. Since we find that a majority of climatic cooling events correlate either robustly or possi bly with coeval LIPs, we cannot reject the hypothesis that LIP emplacement promotes planetary cooling; or stated conversely, we can at least reject the hypothesis that there is no correlation between glaciations and LIPs.
We readily admit that there are other viable approaches to testing this hypothesis. For example, Park et al. (Chapter 7 this volume) test the same hypothesis, but choose to incor porate additional theoretical considerations. Surely many theoretical constraints, such as the paleolatitude of LIPs, are likely important given what we know about the climatic importance of LIPs emplaced during most recent Cretaceous and Cenozoic times (Kent & Muttoni, 2008;2013). However, since even further complications such as true polar wander, e.g., the ~30° Mesozoic "monster shift" (Kent et al., 2015;Fu & Kent, 2018;Fu et al., 2020), are not accounted for in the paleolatitude estimates of Park et al. (Chapter 7 this volume), we argue that additional consid erations such as paleolatitude are potentially ambiguous Note: See referencing in the text, apart from Paleoproterozoic portion, which is after Gumsley et al. (2017) and Ernst and Youbi (2017). at best. There are also concerns about the concept of a "LIP half-life" that are complicated to execute, particularly the choice to exclude LIPs that might have been initially buried, for example, even buried LIPs can later be weath ered if erosion is deep enough (Mitchell et al., 2019). Again, incorporating additional theoretical constraints such as paleolatitude and half-life is surely ideal, but quite compli cated in practice. Our own approach is therefore more focused on the empirical testing of the putative relation ship, and we do so by offering side-by-side timelines of cli mate events and LIPs over the past 720 Myr.

Mechanisms for Global Cooling via LIPs
Earth goes through periods of global cooling (Table 8.1) that can include global, near global, or regional glacia tions, which are observed in the Archean, Paleoproterozoic, Neoproterozoic, Cambrian, Ordovician, Permo-Carboniferous, Eocene-Oligocene, Eocene to middle Miocene, and Quaternary times (e.g., Evans et al., 1997;Augustin et al., 2004;Eyles, 2008;Stern et al., 2008;Hoffman, 2009;Cather et al., 2009;Bradley, 2011;Prave et al., 2016b;Gumsley et al., 2017;. There is an extensive literature on global and regional gla ciations and a variety of causes have been considered (e.g., Raymo, 1991;Berner, 2004, and references therein). For instance, glaciations have been linked to silicate weather ing (and CO 2 drawdown) during major orogenic episodes such as the formation of the Himalayas (Cenozoic glacia tion) and the assembly of Pangea (Permo-Carboniferous glaciation). Major land plant innovations (and their abil ity to extract CO 2 and release oxygen) have also been thought to be a significant factor in causing or at least favoring glaciations, for example, for the Ordovician and Permo-Carboniferous glaciations (Lenton et al., 2012;Kidder & Worsley, 2010;Algeo et al., 2016;Boyce & Lee, 2017). In addition, it is also now recognized that LIPs can contribute to global cooling via at least three different mechanisms: (1) LIP input of SO 2 into the atmosphere (and conversion to sulfate aerosols), (2) weathering of LIP units and CO 2 drawdown, and (3) increased oceanic bio logic productivity and resulting increased CO 2 drawdown (SLIPs are particularly important in the latter).

Volcanic Winter
SO 2 is a greenhouse gas and causes warming for days to weeks. But on a longer term, it causes cooling because it forms sunlight blocking sulfate aerosols (e.g., Bond & Wignall, 2014). The sulfur content of a LIP can be relevant to the amount of SO 2 released by a LIP. Callegaro et al. (2014) showed that the CAMP and Deccan LIPs, both strongly linked to extinction events, have basalts with high sulfur content (up to 1,900 ppm) while the less damaging (not associated with mass extinction) Parana-Etendeka LIP has basalts with lower sulfur content (less than 800 ppm). The effects are more dramatic if the gases are injected into the stratosphere via Plinian eruptions. So called supereruptions (Self & Blake, 2008), can cause cooler cli mate (Rampino et al., 1988;Robock, 2004;Self, 2006;Self & Blake, 2008;Stern et al., 2008). The 720 Ma Franklin LIP (and its extension into formerly attached southern Siberia after Ernst et al., 2016), a large part of which was emplaced into an evaporite basin, correlates closely with the start of the Sturtian glaciation (Macdonald et al., 2010), one of the longest and most severe glaciations in Earth history. Injection of sulfate aerosols into the stratosphere has recently been proposed as the mechanistic link between Franklin and the Sturtian snowball glaciation (Macdonald & Wordsworth, 2017). The Franklin LIP event was further more emplaced in the tropics (Denyszyn et al., 2009), ren dering it additionally climatically important (next section).

Weathering and CO 2 Drawdown
Another mechanism for driving cooling on the Earth's surface is related to silicate weathering. Weathering of continental silicates is enhanced under a warmer and (assumed) wetter climate (at lower latitudes), and basaltic volcanic rocks weather about ~5 to 10 times faster than felsic compositions (Goddéris et al., 2003(Goddéris et al., , 2014Dessert et al., 2003;White & Brantley, 1995). Therefore, CO 2 drawdown due to weathering of flood basalts can lead to global cool down and even glaciations (Cox et al., 2016). LIPs emplaced in the tropical weathering belt may have particularly significant climatic effects, as has been dem onstrated for recent Cenozoic and Cretaceous times (Kent & Muttoni, 2008, 2013. The sedimentary record should reflect certain trace ele ment and isotopic compositions if LIPs are being mas sively weathered in contrast to average continental crust. For example, erosion of LIPs (and particularly their flood basalts) should produce geochemical/isotopic char acteristics (radiogenic Nd and unradiogenic Sr and Os), which are typical of a dominantly mafic provenance (e.g., Mills et al., 2014;Cox et al., 2016).

CO 2 Drawdown Due to Increased Oceanic Productivity
A related mechanism is increased CO 2 drawdown due to increased ocean productivity (e.g., Cox et al., 2016). Oceanic productivity can be enhanced by increasing fertilization of the oceans. For instance, due to the apatite content in basalt, its weathering can effectively fertilize the oceans (Horton, 2015;Cox et al., 2018), with P as a key nutrient on geological timescales (Tyrrell, 1999). Iron fertilization is an effective climatic forcing mechanism and can be caused by great volumes of silicic volcanic ash (Cather et al., 2009). Cather et al. (2009) further conclude that most Phanerozoic cool-climate episodes were coeval with major explosive vol canism in silicic LIPs, suggesting a common link between these phenomena, which they term the icehouse-SLIP  Cather et al., 2009, modified). SLIPs (red) are modified from Bryan (2007) and represent silicic volcanic provinces with documented volumes >10 5 km 3 . Numbers in white boxes are minimum eruptive volumes in millions of cubic kilometers (Bryan, 2007); major silicic volcanic episodes with uncertain eruptive volumes are queried. LIPs (green, range of Ar-Ar and U-Pb ages and yellow, main pulses) are modified from Ernst (2014) and Ernst and Youbi (2017). The age range of Ordovician-Silurian episode of major explosive volcanism is from Huff et al. (1998) and Huff et al. (2000), and that of Late Devonian-early Carboniferous silicic volcanism in Australia is from Bryan et al. (2004). Purple bands are peak pulses of volcanism showing age ranges (Ma) from Bryan (2007), Pankhurst et al. (2000), Huff et al. (1992), and Min et al. (2001). Note that temporally overlapping episodes of major silicic volcanism in northern Europe (ca. 300-280 Ma ago; Neumann et al., 2004), in the Parana-Etendeka province (ca. 133-128 Ma ago; Peate, 1997), and in Afro-Arabia (ca. 30-28 Ma ago; Ukstins Peate et al., 2003) are omitted for clarity. Permian ignimbrite volcanism in South America (López-Gamundí et al., 1994;Breitkreuz & Van Schmus, 1996) is not plotted for lack of adequate age and volume information. See Figure 2 of Cather et al. (2009) for details of Cenozoic ignimbrite flare-up (IFU) volcanism. Light pink and light purple columns allow visual comparison between volcanic episodes and cold paleoclimate intervals. Timing and paleolatitudinal distribution of glaciogenic detritus and other features (blue) and peak glacial intervals (dark blue) are from Frakes and Francis (1988), Frakes et al. (1992), Crowell (1999), Crowley (2000), Isbell et al. (2003a,b), Brenchley et al. (1994), Saltzman and Young (2005), Cherns and Wheeley (2007), Grahn and Caputo (1992), Dromart et al. (2003), Pirrie et al. (1995), Alley and Frakes (2003), Gröcke et al. (2005), and Zachos et al. (2001). Possible short-lived Late Cretaceous-Eocene glacial events in Antarctica (e.g., Miller et al., 2005;Bornemann et al., 2008) are not depicted. Mean tropical sea-surface temperature (black line) has been detrended and smoothed using a 50 Ma window stepping at 10 Ma increments (Veizer et al., 2000) but has not been corrected for pH of seawater (see Royer et al., 2004). Timescale is from Gradstein et al. (2004). hypothesis ( Fig. 8.1). There is also a potential link with LIPs and black-shale-forming oceanic anoxia events (OAEs) (Turgeon & Creaser, 2008;Zhang et al., 2018).

How LIPs Can Cause the End of a Glaciation
Once an ice age is established, planetary albedo increases because snow and ice cover reflects more energy back into space, which would tend to preserve the colder climate (Willeit & Ganopolski, 2018). Therefore, CO 2 emissions from volcanism would be key to taking the planet out of a global ice age. An alternative mechanism is that the low background (i.e., non-LIP-related) vol canic flux on Earth could continue to cause an accumula tion of CO 2 to reach a tipping point, and so a LIP event is not required to end an ice age. Nonetheless, a LIP event that quickly releases large amounts of CO 2 into the atmosphere could cause the sudden end of the glaciation. Multi-eruption modeling of the CO 2 emissions associ ated with the 66 Ma Deccan Traps, for example, suggests plausibly large enough values to have contributed to the warming observed across the Cretaceous/Paleogene mass extinction (Tobin et al., 2017).

POSSIBLE LINKS BETWEEN GLACIATIONS AND LIPS
Here we step through the main glaciations since the Neoproterozoic and consider LIPs that are temporally c orrelated (or not). See summary in Table 8.1. Some cor relations have been previously noted and others are pro posed here for the first time. We grade each glaciation-LIP correlation as either "robust," "possible," or "no known correlation," More precise dating of both LIPs and gla ciations is required to properly assess the proposed age correlations.
The start of the Sturtian glaciation has been matched to the timing of the 720 Ma Franklin LIP of northern Canada (Macdonald et al., 2010) and coeval Irkutsk LIP in formerly attached southern Siberia (Ernst & Bleeker, 2010;Ernst et al., 2016). Both the low paleolatitude and the emplacement into an evaporite basin might have ren dered the Franklin-Irkutsk LIP climatically important for abetting the onset of the Sturtian glaciation due to enhanced weathering and sulfate aerosol injection into the stratosphere, respectively (Macdonald & Wordsworth, 2017). Additional magmatism of this age is also present in the Kalahari craton (Mutare swarm) and formerly attached Dronning Maud Land region of Antarctica (Fingeren) (Gumsley et al., 2019).
There is, however, no known LIP associated with the end of the Sturtian glaciation ca. 660 Ma, and it therefore is possible that the termination of the glaciation was due to ambient buildup of CO 2 from a continuous back ground level of global volcanism.

Marinoan Glaciation (ca. 640-635 Ma)
The other important Neoproterozoic glaciation is the ca. 640-635 Ma Marinoan glaciation (  Hoffman et al., 1998;Schrag et al., 2002;Prave et al., 2016b). Its start is not well defined, and the size of the gap with the end of the prior 720-660 Ma Sturtian glacia tion is also poorly known. Like the Sturtian glaciation, diamictites of the Marinoan glaciation are terminated by transgressive sequences and postglacial cap carbonate units, which provide evidence for a strong CO 2 hysteresis leading to extreme weathering during the postglacial supergreenhouse (Bao et al., 2008;Hoffman et al., 1998;Hoffman et al., 2017).
We speculate on a potential climatic link with the newly recognized ca. 650-630 Ma Wudang dyke swarm of South China (Zhao & Asimow, 2018). Numerous ca. 650-630 Ma mafic and ultramafic dykes intruded the Neoproterozoic Wudang Group in the northern Yangtze Block, South China (Wudang mafic dykes). U-Pb zircon ages from the Wudang mafic dykes span from ca. 651-627 Ma and their petrogenesis argues that the dykes are most likely interpreted as a LIP and not related back-arc extension (Zhao & Asimow, 2018). More precise U-Pb geochronology is needed to determine a link with either the start or end of the Marinoan glaciation.

The Gaskiers Glaciation (579 Ma)
Following the two presumed snowball glaciations (Sturtian and Marinoan), Ediacaran glacial deposits have been found on nine different continents (Hoffman & Li, 2009; see also McGee et al., 2015). Assuming synchrone ity, this 579 Ma Ediacaran event referred to as the Gaskiers glaciation was likely too short (<340,000 years) to have been a multimillion-year-long snowball glaciation (Pu et al., 2016). Except for those in low-latitude Australia, Ediacaran glacial deposits are found on mid latitude or high-latitude continents, similarly inconsistent with a snowball Earth origin (Hoffman & Li, 2009).
The age of the Ediacaran Gaskiers glaciation is r emarkably similar to the middle, most extensive   Ogg et al., 2016). The carbon-isotope curve is a smoothed version modified by Ogg et al. (2016) from the synthesis for the late Proterozoic by Cohen and Macdonald (2015) calibrated by them to the Cryogenian timescale of Rooney et al. (2015). Ranges and images of organic-walled microfossils, Ediacaran metazoans, and bioturbation styles are from Narbonne et al. (2012). Timing of events associated with continental collision and convergent and divergent plate margin activities related to assembly and dispersal of Rodinia and subsequent assembly of Gondwana is from Cawood et al. (2016). Timing of large igneous provinces (LIPs) is from Ernst (2014), Ernst and Youbi (2017), Macdonald and Wordsworth (2017). Additional geochemical trends, biostratigraphic ranges, regional stages, and details on calibrations are compiled in Shields-Zhou et al. ca. 590-570 Ma LIP pulse within the more protracted 615-555 Ma multipulsed Central Iapetus Magmatic Province (CIMP) (e.g., Puffer, 2002;Ernst & Bell, 2010;Ernst & Bleeker, 2010;Youbi et al., 2011Youbi et al., , 2018Youbi et al., , 2019Youbi et al., , 2020Mitchell et al., 2011;Larsen et al., 2018;Tegner et al., 2019;Kjøll et al., 2019). This pulse includes the Grenville dykes of eastern Laurentia (ca. 590 Ma), the Volyn flood basalts of Baltica at ca. 570 Ma (Shumlyanskyy et al., 2016), and particularly the ca. 580 Ma Ouarzazate intraplate magmatism of West Africa (Youbi et al., 2011). Paleomagnetic comparison demon strates that these three continents were contiguous ca. 615 Ma (Robert et al., 2018), supporting the interpreta tion of a shared CIMP LIP, until the opening of the Iapetus Ocean (Robert et al., 2018;Mitchell et al., 2011). The Ouarzazate group of Morocco includes both a SLIP f ollowed by a LIP (i.e., a pulse of silicic volcanism f ollowed by pulse of basaltic volcanism), both ca. 580 Ma, and we tentatively suggest that the SLIP caused the glaciation and the LIP (flood basalts) ended the shortlived glaciation Youbi et al., 2011Youbi et al., , 2018Youbi et al., , 2019Youbi et al., , 2020. Other portions of CIMP in Laurentia and Baltica during this ca. 580 Ma time interval spanning the Gaskiers glaciation (consisting of mainly mafic mag matism), also need to be evaluated to determine the pre cise timing and nature of their climatic contributions, whether warming (through CO 2 release) or cooling (through weathering and CO 2 drawdown).
The 540-530 Ma Wichita LIP is exposed in the Ouchita aulacogen of the southern United States, but is covered by younger rocks outside of it. Geology and geophysics reveal suites such as the Roosevelt gabbros, the Glen Mountains mafic-ultramafic layered intrusion, and the Navajo Mountain basalt-spilite group (variably altered basaltic to intermediate volcanic rocks), as well as a significant vol ume of associated A-type silicic magmatism consisting of Wichita granites and Carleton rhyolites (e.g., Hanson et al., 2013). Given its 540-530 Ma age, the Wichita LIP can be tentatively linked to the cause of the Lower Cambrian glaciation (ca. 535 Ma) in Avalonia and Baltica.
The ca. 511 Ma Kalkarlindji LIP is widespread in Australia as basaltic remnants of a much more extensive flood basalt province that also includes intrusive compo nents (Jourdan et al., 2014;Marshall et al., 2018, and Chapter 19 this volume, and references therein).
Magmatism likely began at around 512 Ma, with final eruption occurring between 509 and 498 Ma. The Kalkarindji LIP (Australia) shows synchrony with the Early and Middle Cambrian (Stage 4-5) extinction and could have also be linked to the possible cooling period and/or glaciation that occurred during the Middle to Late Cambrian, between ca. 513 and ca. 488 Ma.

The Hirnantian Glaciation (ca. 440 Ma)
The end-Ordovician mass extinction consists of two pulses, a base Hirnantian (end-Katian) pulse and a late Hirnantian pulse, with ages of ca. 445.2 Ma and ca. 443.8 Ma, respectively (Ogg et al., 2016). The former is associated with the onset of global cooling and associated sea level drop (nearly 100 m) and the latter is associated with a global rewarming, sea level rise and widespread anoxia (Harper et al., 2014). Furthermore, the glaciation is preceded by a period of warming in the late Katian known as the Boda event (Fortey & Cocks, 2005;Lefebvre et al., 2010).
There are indications of magmatism being involved in the end-Ordovician extinction. Buggisch et al. (2010) concluded the presence of major volcanism, on the basis of the prominent Deicke, Millbrig K-bentonite beds of North America and the Kinnekulle K-bentonite bed of Europe, both related to SLIP (i.e., supereruptions, Fig. 8.1; Cather et al., 2009), the latter precisely dated at 454.52 +/-0.50 Ma (U-Pb zircon; Svensen et al., 2015). Recently, five of the K-bentonites have been dated by high-precision chemical abrasion-thermal ionization mass spectrometry (CA-TIMS) U-Pb zircon geochronol ogy, where the Kinnekulle K-bentonite gives an age of 454.06 ± 0.43 Ma. The entire volcanic activity recorded in the Ordovician at Sinsen section in Olslo has a duration of 4.05 ± 0.91 Myr (Ballo et al., 2019).
Hg values in marine strata exposed at the Wangjiawan Riverside section in South China show a significant peak that links the Hirnantian glacial maximum and the Late Ordovician (ca. 444 Ma) mass extinction directly to mafic volcanism (Jones et al., 2017). LIP candidates of approxi mately the correct age have been suggested (e.g., Kravchinsky, 2012;Khudoley et al., 2013;Retallack, 2015). These include the ca. 450 Ma (457 ± 34 Ma -379 ± 27 Ma) Suordakh dolerite event in eastern Siberia (Khudoley et al., 2013;Chamberlain et al., 2018), the Ongnyeobong Formation volcanics in South Korea (Cho et al., 2014), flood basalts of Sierra del Tigre in Argentina (González-Menéndez et al., 2013;Retallack, 2015), and the intraplate basaltic lava flows of the Soltan Maidan Basaltic Complex with thickness up to about 1,300 m located in the eastern Alborz zone, north of Iran (Derakhshi et al., 2017). More precise and better age determinations are required to test the link to the Hirnantian glacial maximum and Late Ordovician mass extinction. Nonetheless, Hg concentrations in marine strata are known to be associated with well-studied LIPs, such as the Siberian Traps, the CAMP, and the Deccan Traps, suggesting that their occurrence immediately before the Hirnatian glacial maximum provides some of the first direct evidence linking LIP emplacement with climatic cooling (Jones et al., 2017, and references therein).
Other causes have been proposed for the end-Ordovician extinction. Widespread glaciations have been linked to silicate weathering during major orogenic episodes (e.g., Raymo, 1991;Berner, 2004). Major land plant innova tions have also been thought to be a significant factor in causing or at least favoring glaciations. Lenton et al. (2012) applied this observation specifically to the end-Ordovician glaciation, as did Kidder and Worsley (2010).

Devonian Glaciations
Widespread glaciation occurred in Gondwana in the Late Famennian, at the end of the Late Devonian (e.g., McGhee, 2014a,b). Evidence in Gondwana include biostratigraphically dated glacial tillites, glacial striated pavements and clasts, and ice-rafted dropstones where continental ice sheets covered at least 16 × 10 6 km 2 of land (Lopez-Gumundi & Buatois, 2010). In Laurussia, there were also Late Famennian glacial tillites produced by lowland glaciers as close as 30°S to the equator. This The Yakutsk-Vilyui LIP is widespread in the Siberian craton and dominantly consists of a giant radiating mafic dyke swarm and radiating rift system converging to a m antle plume center on the eastern side of the craton (Kiselev et al., 2012). The largest volumes of flood basalts are preserved in the Vilyui-Marka rift system. U-Pb geo chronology indicates emplacement in two pulses, at ca. 374 Ma and ca. 363.4 Ma, that have been linked with mass extinctions (Ricci et al., 2013;Polyansky et al., 2017Polyansky et al., , 2018Ernst et al., 2019). There are also associated car bonatites and the diamondiferous kimberlites. The pos sible environmental impact of carbonatites and associated alkaline magmatism has been discussed by Ray and Pande (1999).
The ca. 370 Ma Kola-Dnieper LIP of Baltica (also known as the East European craton) includes coeval mafic magma tism of the Dnieper-Donets rift, a dyke swarm extending for 2,000 km along the Urals, and magmatism in the Kola pen insula and elsewhere in the Baltic craton (Nikishin et al., 1996;Kravchinsky, 2012;Puchkov et al., 2016). Two plume centers are recognized (Puckhov et al., 2016). Kimberlites (Archangelsk) and carbonatites (Kola Alkaline province) are also associated with this event.
To date, a range of chronometers has been applied to determine eruption ages from across the region of the EUNWA group including whole-rock Rb-Sr and K-Ar dating; 40 Ar/ 39 Ar dating of mineral separates; and U-Pb dating of zircon, titanite, and perovskite (e.g., Timmerman et al., 2009). The duration of activity is currently estimated to span a period of ca. 100 million years, from Early Carboniferous to Upper Permian-Early Triassic (350-250 Ma), with several hiatuses . Three main pulses can be distinguished at ca. 300 Ma, 290-275 Ma, and 250 Ma, and each of these pulses can be considered a separate LIP within the overall EUNWA group.
The distinct 300 Ma pulse of this multipulsed LIP, Skagerrak-Centered Large Igneous Province (the "SCLIP"; Torsvik et al., 2008; see also Ernst & Buchan, 1997) can be considered plume related. The start of this glacial time period (associated with P1) coincides with a particularly well-defined portion of this EUNWA event, the 300 Ma SCLIP pulse that includes magmatism (vol canics, sills, and dykes) in the UK, Norway, Sweden, and in parts of the intervening North Sea, and including a giant radiating dyke swarm focused at the southern end of the Oslo rift (Ernst & Buchan 1997;Torsvik et al., 2008).
The EUNWA event(s) is well represented in Morocco in northwestern Africa and also in southern Scandinavia and northern Germany. The huge volume of extruded and intruded magmatic products of the EUNWA event(s) (example in the Oslo Graben, the estimated volume is at ca. 35,000 km 3 while in the North German Basin, the total volume of felsic volcanic rocks, mainly rhyolites and rhyodacites, was of the order of 48,000 km3) has led to suggestions of a thermally anomalous mantle plume to explain this pulse of the magmatism (Ernst & Buchan, 1997;Torsvik et al., 2008). A significant detractor from a plume hypothesis is the duration of activity and the helium isotope signature of lithospheric mantle xenoliths from the Scottish Permo-Carboniferous dykes, sills, and vents . While the EUNWA event(s) qualifies as a significantly large volume of magmatism, there is some uncertainty whether it technically qualifies as a LIP pending better geochronology.
The EUNWA event(s) may have contributed to the great Gondwanan glaciation that occurred from the Late Devonian to the Late Permian (Veevers & Powell, 1987;Crowell, 1999;Isbell et al., 2003a,b). Glaciers achieved their maximum paleolatitudinal range between the mid dle Stephanian (ca. 305 Ma ago) and near the end of the Sakmarian (ca. 284 Ma ago) (Isbell et al., 2003a,b). This hypothesis is termed the "icehouse-silicic large igneous province (SLIP)" hypothesis (Cather et al., 2009).
The P2 and P3 glaciations (ca. 290 and ca. 270 Ma, respectively) can be linked with the second and the third pulses of the EUNWA event and the Tarim LIP, which is widespread in Central Asia (Xu et al., 2014) and the poorly dated but also ca. 290 Ma Panjal-Qiangtang LIP (Zhai et al., 2013;Shellnut, 2016), and potentially with the 300-280 Ma Tianshan LIP (Kravchinsky 2010). The P4 glaciation (ca. 260 Ma) approximately matches with the 260 Ma Emeishan LIP of South China (e.g., Shellnut, 2014). The Choiyoi silicic province (possible SLIP) with a peak age of 265 Ma may also have had a role in P3 (Kimbrough, 2016a,b).
For Early Carboniferous time, the Jebilets, Rehamna, and Fourhal basins of western Meseta (Central Hercynian Massif of Morocco) show compelling similarities in tec tonosedimentary evolution. Their deposits record large instabilities and disorganization with huge thickness and lithological variations related to synsedimentary tecton ics. At the same time, tectonic tilting affects the basement of these basins, with blocks controlled by bordering transfer faults. Abundant traces of magmatic activity during the Carboniferous period are recognized in Morocco, particularly in the Jebilets, Rehamna, and Fourhal areas. These rocks constitute a magmatic prov ince (called here the Moroccan Meseta LIP) consisting of basaltic lavas, mafic sills and dykes, and gabbroic intru sions, together with subordinate layered ultramafic intru sions and silicic volcanic/intrusive rocks exposed in the Meseta Domain as part of the Moroccan Variscan belt.  Available zircon U-Pb ages obtained  from various rocks in this province, which has an areal extent of ~400,000 km 2 (~850 km × 470 km), indicate that magmatism occurring between 349 and 330 Ma ago with peak at 340, coeval with the eastern Meseta volcanism in northeastern Morocco. This LIP could be the first pulse of the EUNWA province. Further geochronology work is required to determine whether this ca. 340 Ma age, some 40 Myr older than the ca. 300 Ma SCLIP, represents a new intraplate event or belong to a long-lived multipulsed 359-250 Ma EUNWA LIP. Although possibly compelling correlations for LIPs with Permo-Carboniferous glaciations, we are compelled to point out that Gondwana was also tectonically transit ing the South Pole at this time (Caputo & Crowell, 1985). Like Antarctica today, having a landmass transit the polar regions is likely to promote growth of continental ice sheets, so it is more difficult to distinguish the effects from LIPs and high paleolatitude during the Permo-Carboniferous glaciations.

Jurassic Glaciations
The Jurassic (~201-145 Ma) was long considered a warm greenhouse period. However, a recent study has suggested that cool, even icehouse episodes occurred (Korte et al, 2015). Oxygen-isotope trends and other proxies suggest a general warming from Hettangian through Sinemurian, followed by a cool Late Pliensbachian, and a warm to very warm Toarcian, then a cooling Aalenian through Bajocian (e.g., Korte & Hesselbo, 2011;Korte et al., 2015). In addi tion, the generally warm climate of Callovian through Kimmeridgian was interrupted by an anomalous late Callovian cold period (e.g., Dromart et al., 2003;Pellenard et al., 2014). The Tithonian is generally interpreted as cooler and more arid in many regions. However, most of these carbon-isotope, oxygen-isotope, and temperature trends are derived from individual basins in Western Europe, and a verified global synthesis has not yet been compiled (e.g., see discussions in Wierzbowski, 2015).
The precisely dated 183 Ma Karoo-Ferrar LIP of southern Africa and adjacent Antractica (Hastie et al., 2014;Burgess et al., 2015a) marks the end of the Pliensbachian and therefore likely caused the end of the Late Pliensbachian cooling event, but there is no known LIP event earlier than ca. 183 Ma that would explain the Late Pliensbachian cooling.
The Aalenian (174.1 Ma to about 170.3 Ma) through Bajocian (170.3 Ma to around 168.3 Ma) cooling is not associated with any known LIP activity, except that a pulse of the Karoo LIP is thought to be younger (178 Ma) on the basis of Ar-Ar dating (Jourdan et al., 2008). However, Svensen et al. (2012) concludes "there are no indications of volcanism younger than 181.8 [Ma] even when considering the full range of the 2-sigma uncertain ties." Another way to cause the ca. 174 Ma cooling would be to hypothesize that erosion of widespread Karoo flood basalts (and resulting CO 2 drawdown) occurred about 8 Myr later after emplacement at 183 Ma. The Late Callovian (ending at 166.1 ± 4.0 Ma) does not have any known LIP link.

Cretaceous Glaciations
The Cretaceous (~145-66 Ma) was long considered a warm greenhouse period. However, there is an abundance of evidence for short intervals of cold climatic conditions during the Early Cretaceous (late Valanginian-earliest Hauterivian, late early Aptian, latest Aptian-earliest Albian; Bodin et al., 2015). Each of these intervals is associated with rapid and high amplitude sea level fluc tuations, supporting the hypothesis of transient growth of polar ice caps. As evidenced by positive carbon isotope excursions, each cold episode is associated with enhanced burial of organic matter on a global scale.
From a LIPs perspective, there is a relatively good match between the timing and size of LIP eruptions and the amplitude of Early Cretaceous warming episodes (Bodin et al., 2015). In addition, with a slight lag of a few million years in some cases, there is the onset of a cooling. For instance, the late Valanginian-earliest Hauterivian cool ing (136-134 Ma, Fig. 3 in Bodin et al., 2015) would cor respond to a number of events of age ca. 135 Ma (cf., Bryan & Ferriera, 2013), precisely dated 135-133 Ma Parana-Etendeka (Pinto et al., 2011;Almeida et al., 2018;Hartmann et al., 2019), initial 134 Ma Comei-Bunbury pulse of Kerguelen LIP (Zhu et al., 2009), and ca. 130 Trap LIP event of southwest Greenland). The late early Aptian cooling (124-123 Ma; Fig. 3 in Bodin et al., 2015) could be associated with the precise U-Pb dating of the initial 126-121 Ma pulse of HALIP (ages mainly from Corfu et al., 2013; see additional ages and summaries in Kingsbury et al., 2018;Dockman et al., 2018) and the lat est Aptian earliest Albian (117-112 Ma; Fig. 3 in Bodin et al., 2015) could be associated with stages of the Kerguelen oceanic plateau and also the Rajmahal Traps of eastern India (Fig. 8.4).
The interval following the late Turonian through Maastrichtian (ca. 90-66 Ma) is considered to have been a period of significant global cooling, possibly driven by a combination of declining pCO 2 levels and opening ocean gateways (Cramer et al., 2011;Linnert et al., 2014). From a LIPs perspective, this timing starts with three impressive ca. 90-95 Ma LIP events: Caribbean-Colombian, Madagascar, and the second 95 Ma pulse of the High Arctic LIP (HALIP) (Kingsbury et al., 2018), and therefore the possibility of any or all three contribut ing to global cooling remains plausible. Furthermore,  Bodin et al., 2015, modified). LIPs and SLIPs records are modified from Ernst (2014) and Ernst and Youbi (2017). Note the links between LIPs and the climatic state (H = Hothouse, G = Greenhouse, and C = Coldhouse) as already noted by Bodin et al. (2015).
this cooling trend post-LIP emplacement may have con tinued as is observed due to the long-term effect of car bon burial associated with OAE2 reducing CO 2 (Berner, 2006), which itself may have been triggered by these aforementioned LIPs (Turgeon & Creaser, 2008). The end of this interval would correlate in timing with the 66 Ma Deccan LIP, which has been modeled to have driven sig nificant climatic warming (Tobin et al., 2017).

Cenozoic Glaciations
The Paleocene and Eocene epochs of the early Cenozoic (66-33 Ma) are regarded as greenhouse climates on account of the highly elevated concentrations of CO 2 (>750 ppm; Zachos et al., 2001). Following this peak warmth of the Cenozoic, however, marine oxygen isotope records indicate a global cooling trend that appears to have culminated rapidly at the Eocene/Oligocene bound ary 33.9 Ma (Zachos et al., 2001;Mudelsee et al., 2014;Fig. 8.5a). The first continental-scale glaciation of Antarctica occurred in the earliest Oligocene epoch (33.9 Ma), followed by the onset of northern-hemispheric gla cial cycles in the late Pliocene epoch about 31 Myr later (Zachos et al., 2001).
The ca. 30 Ma Afro Arabian LIP is associated with opening of the Red Sea and Gulf of Aden and extension in the East African rift system. Precise dates indicate a main pulse of LIP magmatism at 31-29 Ma (e.g., Ukstins et al., 2002), which would be a few million years too young to be the cause of the 33.9 Ma onset of the continental    (Zachos et al., 2001). LIPs and SLIPs records are modified from Ernst (2014) and Ernst and Youbi (2017). Ignimbrite flux estimates for individual volcanic fields (colored lines) are derived from the age data and volume estimates of Lipman (2000), McIntosh et al. (1992), Best and Christiansen (1991), McIntosh and Bryan (2000), McDowell (2007), McDowell and McIntosh (2007), Ferrari et al. (2002), Chapin et al. (2004), and Moye et al. (1988). Black dashed line is the overall eruptive flux for the IFU. The atmospheric flux of elutriated volcanic ash during the IFU is assumed to be subequal to the ignimbrite eruptive flux (after Zachos et al., 2001, andCather et al., 2009, modified).
scale g laciation in Antarctica (the Oi2 global cooling event). However, Prave et al. (2016a) obtained geological and Ar-Ar and U-Pb geochronological data that define four pulses of volcanism for the Lake Tana region in the northern Ethiopian portion of the Afro-Arabian LIP, the oldest of which is a ~1-km-thick flood basalt likely as old as ~34 Ma (40Ar/39Ar age of 34.05 ± 0.54/0.56 Ma), but of unknown duration. This 34 Ma age is older than the 31-29 Ma ages typically attributed to Ethiopian flood basalts and could be responsible for the 33.9 Ma global cooling event. The felsic volcanism was the product of super eruptions that created a 60-80 km diameter caldera marked by km-scale caldera-collapse fault blocks and a steep-sided basin filled with a minimum of 180 m of s ediment and the present-day Lake Tana (Prave et al., 2016a) (Fig. 8.5). CO 2 decline in the late Eocene has been implicated to drive glaciation on Antarctica (DeConto & Pollard, 2003), and we argue that the Afro-Arabian LIP provided the driver for this CO 2 drawdown.

DISCUSSION AND CONCLUSIONS
1. LIPs and SLIPs magmatism can have significant global climatic effects including causing mass extinction events via a complex web of changes (characteristically rapid) in atmospheric/oceanic acidification, oceanic anoxia, sea level, toxic metal input (e.g., mercury: Hg), and most significantly in temperature, both warming and cool ing. In the broadest sense, LIPs and SLIPs can affect (or even induce) shifts between icehouse, greenhouse, and hot house climatic states (e.g., Kidder & Worsley 2010, 2012. 2. Earth has gone through periods of cooling including global, near global, or regional glaciations, which are observed in the Archean, Paleoproterozoic, Neopro terozoic, Ordovician, Permo-Carboniferous, and Cenozoic times. We have reviewed the ways in which LIPs and SLIPs can cause global cooling: We focus on glacia tions of Neoproterozoic and Phanerozoic age to identify potentially age-correlated LIP events. The 720 Ma Franklin-Irkutsk LIP is linked with the start of the Sturtian glaciation and the Wudang dyke swarm (ca. 650-630 Ma) is a LIP that spans the Marinoan glaciation. The ca. 580 Ma pulse of the Central Iapetus Magmatic Province (CIMP) is associated with the start and end of the 579 Ma Gaskiers glaciation. The Hirnantian glacia tion (ca. 440 Ma) may be associated with poorly dated ca. 440 Ma intraplate magmatism in several regions, includ ing eastern Siberia, South Korea, Argentina, and else where. Permo-Carboniferous glaciations (P1-P4, 300-260 Ma) can be correlated with widespread intraplate magma tism of the European northwest African magmatic prov ince (and its initiation as the 300 Ma Skagerrak LIP), and also the 260 Ma Emeishan LIP of China. The ca. 30 Ma initial pulse of the Afro-Arabian LIP approximately matches the Eocene-Oligocene cooling. More precise dat ing of both the LIP and cooling events is required to con firm the correlations and to assess the role of LIPs relative to other causes of global cooling.
3. A variety of causes have been considered for global and regional glaciations. For instance, glaciations have been linked to silicate weathering (and CO 2 drawdown) during major orogenic episodes such as the formation of the Himalayas (Cenozoic glaciation) and the assembly of Pangea (Permo-Carboniferous glaciation). Major land plant innovations (and their ability to extract CO 2 and release oxygen) have also been thought to be a significant factor in causing or at least favoring glaciations (e.g., for the Ordovician and Permo-Carboniferous glaciations). In addition, it is also now recognized that LIPs can pos sibly contribute to global cooling via at least three differ ent mechanisms: (1) LIP input of SO 2 into the atmosphere (and conversion to sulfate aerosols), (2) weathering of LIP units and CO 2 drawdown, and (3) increased oceanic biologic productivity and resulting increased CO 2 draw down (SLIPs are particularly important in the latter).
4. There are many LIP events that are not linked to shifts to hotter or colder conditions, and so it could be inferred that environmental conditions at the time of LIP emplacement must already be close to a tipping point such that the LIP provides the final critical threshold that causes a rapid climatic shift. In the present context, we could view the non-LIP factors such as supercontinent breakup and its palaeogeography during rifting and con tinental arc (McKenzie et al., 2016;Macdonald et al., 2019) and submarine volcanism as external forces to have the effect of preparing the climate for a superimposed shock from LIP magmatism (silicic initially, and then mafic). These other factors are generally of longer dura tion and so the changes they would introduce would also be of longer duration. In contrast, a short-duration LIP pulse could cause a sudden change in climate.

ACKNOWLEDGMENTS
Richard E Ernst and Nasrrddine Youbi are supported by Russian MegaGrant 14.Y26.31.0012, and Richard E Ernst is also supported by Canadian NSERC CRD grant NSERC/CRD 52313117. Funding was also provided by the Academy Hassan II of Science and Technology of Morocco AcadHIIST/SDU/2016-02 (grant to Abder razzak El Albani, Nasrrddine Youbi, and others), the Swedish Research Council (2015-05875) (grant to Ulf Söderlund, Nasrrddine Youbi, and Richard E. Ernst), the agreement relating to scientific and technical cooperation between the University of São Paulo and the Cadi Ayyad University (to Colombo Celso Gaeta Tassinari and Nasrrddine Youbi), and technical cooperation between CNRST (Morocco, SU 01/ 2019-2020; to Nasrrddine Youbi) and FCT (Portugal, Biennial Programme 2019-2020; to João Mata and José Madeira). This is a contri bution to IGCP 648. We appreciate the two thoughtful reviews by two anonymous reviewers.