August 2013 LIP of the Month

Large Igneous Provinces Trigger Hothouse Climate

David L. Kidder
(email: kidder@ohio.edu)

Thomas R. Worsley^[1]
(email: tomworsley@aol.com)

Department of Geological Sciences
Ohio University
Athens, OH  45701
U.S.A.

Introduction

When Alfred Fischer proposed that Earth alternated between icehouse and greenhouse states through Phanerozoic history (Fischer, 1981), he suggested that changes between these states were strongly influenced by volcanic release of carbon dioxide to cause climate warming, and that cooling could be driven by withdrawal of this atmospheric greenhouse gas as rocks weathered and carbon was buried during sedimentation.  Fischer’s vantage point came from examining the realm of ancient oceans during unglaciated intervals.  Sea-floor records revealed significant differences relative to modern oceans.  For example, Emiliani (1953) determined that Mesozoic deep oceans were quite warm (~15° C), and numerous studies supported these initial findings (see Hay, 2013 for review).  Ocean sediment cores confirmed and augmented on-land observations that some ancient oceans were marked by widespread distribution of anoxic sediments.  The paleoceanographers realized that widespread ocean anoxia was associated with warming temperatures, rising sea levels and positive δ¹³C anomalies that indicated extensive organic carbon burial.  The pulse-like nature of these occurrences led to the description of these intervals as Oceanic Anoxic Events (OAEs).  The OAE helped explain the systemic linkages in the ocean system that bring about such coordinated change (e.g. Schlanger et al., 1976; Fischer and Arthur, 1977; Jenkyns, 1980; Leckie et al., 2002). 

With the increase in global paleoclimate studies in recent decades, terms like supergreenhouse, hothouse, and hyperthermal event arose in the literature implying that there are intervals in Phanerozoic history that are warmer than a typical greenhouse climate.  Our synthetic analysis of a particularly warm interval associated with the end-Permian extinction (Kidder and Worsley, 2004) led us to introduce descriptive and functional criteria that help distinguish icehouse, greenhouse, and hothouse climates.  We suggested that a Mid-Permian climate shift from icehouse to greenhouse may have been triggered by reductions in silicate weathering and associated carbon burial as orogenically-driven weathering of the large and extensive mountains formed by Pangean assembly declined (Kidder and Worsley, 2004).  Computer modeling by Kiehl and Shields (2005) suggests a weakening of pole-driven thermohaline circulation in the Late Permian greenhouse climate. 

The LIP link to warming climates

We further suggested that the Siberian Traps Large Igneous Province (LIP) forced the already warm Late Permian greenhouse world into the hothouse state we began to define in 2004.  Later (Kidder and Worsley, 2010), we generalized our Permian-Triassic model to include about a dozen Phanerozoic intervals (Fig. 1; Table 1).  We emphasize that a number of researchers had already begun linking basaltic LIP activity to climate warming and mass extinctions.  Even as Fischer was proposing the icehouse and greenhouse states (Fischer, 1981), Keith (1982) suggested the link between LIPs warming episodes, anoxia, and mass extinctions in Earth history.  Kerr (1998) proposed connections between oceanic LIP activity, warming, and mass extinction at the Cenomanian-Turonian boundary.  Wignall (2001) expanded that thinking to include a number of mass-extinction intervals.  Separately, Courtillot et al. (1986) proposed that Deccan Traps volcanism caused the end-Cretaceous extinction, and Courtillot and Renne (2003) extending that linkage to multiple extinction horizons.  Keller (e.g. 2005; 2008) further emphasized and expanded upon these connections.

The hothouse planetary state

Strong proxy evidence for warm deep oceans as well as polar areas that warmed much more substantially than tropical surface oceans is well recognized in the Mesozoic and evident in much of the Paleozoic (see Hay, 2013 for review).  Heating of this nature requires a mechanism, but computer modeling of this complex change has proven a challenging task, even as technology and understanding have evolved in recent decades.  One difficulty is that warming Earth’s surface to a thermally equable climate mandates increased meridional heat transport to the poles.  However, achieving such equable warming greatly diminishes the main equator-pole heat transport mechanisms (winds and ocean currents) that characterize today’s Earth.  The conundrum could be solved with a reversal of Earth’s modern mode of thermohaline circulation such that sinking low-latitude warm brines drive the deep-ocean density-driven currents rather than the sinking cold brines of the icehouse state, as was first suggested by Chamberlin (1906) and reiterated by Stommel (1961).  Brass et al. (1982) revived the idea yet again as the evidence for warm ancient greenhouse oceans grew increasingly robust.  Zhang et al. (2001) simulated sinking warm ocean brines in a computer model, but their results suggested that achieving this condition was difficult, and that it was unstable and probably unsustainable.  They called this sinking-warm-brine condition the haline mode (Fig. 2).  The contrasting thermal mode of thermohaline deep-ocean circulation is the familiar icehouse condition in which sinking cold brines drive thermohaline ocean circulation. 

HEATT Model and HEATT Episodes

In our recent model (Kidder and Worsley, 2010), we suggested that geologically short hothouse excursions that typically last less than 1 m.y. be named HEATT episodes, where HEATT stands for Haline Euxinic Acidic Thermal Transgression.  We hypothesized that changes in the Earth system should be governed by principles of global heat transport as they relate to planetary windbelts and ocean circulation.  We then synthesized geologic, paleontologic, paleoceanographic, and paleoclimatic data, and integrated those results so as to test theoretical predictions.  For example, atmospheric and oceanic circulation would have to adapt as the thermal contrast between the poles and equator diminished in a warming world.  Emanuel (2002) suggested that significant changes in such factors should yield multiple climate states (he called them regimes).  We were unaware of his work in 2010, but his contributions emphasize the importance of changes in atmospheric and ocean circulation as well as tropical cyclones in driving the manner in which Earth shifts from one climate state to another. 

In our model (Kidder and Worsley, 2010), we assume that a weak thermal mode, defined by sinking of cool, but not frigid polar sinking waters that very sluggishly fill the deep ocean characterizes greenhouse climate.  We suggested that greenhouse conditions (e.g. Fig. 2) characterized about 70% of the Phanerozoic.  We proposed the greenhouse as a default state from which icehouse and hothouse climates can be forced.  Cooling to an icehouse (e.g. via orogenically-intensified silicate weathering and related carbon burial) leads to glaciation and polar sinking of thermal-mode cold brines (see Kidder and Worsley, 2012 for discussion and caveats).  Warming from greenhouse to hothouse is driven by LIP emissions of greenhouse gases.  Under favorable preconditions (i.e. in the absence of significant competing cooling forcing mechanisms), a warm greenhouse climate can be forced to a hothouse state via a LIP trigger.  Other triggers may also be possible (see below), but the key point is that a sufficiently potent forcing mechanism allows the haline mode to temporarily supplant the weak thermal mode of greenhouse climate, delivering warm saline waters to the deep ocean in subtropical to middle latitudes (Fig. 2).  Such delivery will be most effective where restricted basins/embayments provide local intensification of brine formation (e.g. modern Mediterranean/Red Sea analogs).  Already weakening planetary windbelt velocities will cut poleward heat transport via wind-driven surface ocean currents.  Figure 2 illustrates intensified mechanisms for poleward transfer of heat from the tropics via increased latent heat transport in the atmosphere and intensified heat transport by stronger, deeper reaching cyclones that extend to high latitudes.  The haline mode will deliver heat from middle latitudes to polar oceans (Fig. 2) as thermohaline circulation changes direction in hothouse climate.

The reduced planetary windbelt velocities will have additional consequences.  For example, wind-driven upwelling will diminish.  Hay (2008) suggested that equatorial upwelling, though weakened, will likely persist despite overall lessening of planetary windbelt velocities.  The consequences of reduced windbelt-driven upwelling will be declines in nutrient delivery to shallow waters where photosynthesis occurs.  This nutrient shortage will intensify as the weakening planetary windbelts blow less iron-bearing dust into the open ocean.  An iron shortage means that nitrogen-fixing bacteria (diazotrophs) will not produce as much usable nitrogen, so availability of usable nitrogen will be further reduced. Oceans will thus likely become highly N-limiting, with Fe limitation serving as an important driver of loss of available N.  Usable phosphate will likely reside in widespread anoxic waters (e.g. Hotinski et al., 2000), but as productivity drops due to Fe- and N-limitation, such phosphate may not be fully utilized as demand for it diminishes.  The reductions in upwelling noted above will hamper bringing nutrients to the photic zone, however, the increase in cyclonic storm activity and depth of reach may become an important mechanism for delivering dissolved phosphate and iron to surface waters.  This tempts speculation that cyclones will create ephemeral swaths of ocean productivity in greenhouse oceans.  However, as the haline mode intensifies, cyclone-driven upwelling becomes deadly.  The sulfate delivered to widespread zones of anoxic waters will be reduced to sulfide, creating euxinic conditions.  Water column genesis of framboidal pyrite will drain iron from euxinic waters as the pyrite sinks to bottom.  Upwelling in hothouse oceans would be likely to draw up iron-free euxinic waters rather than nutrients, resulting in starving and poisoning of pelagic biotas.

A HEATT episode develops as anoxia expands in a warming greenhouse ocean (Fig. 3).  As the haline mode becomes established, the sinking warm brines heat the ocean sufficiently to instigate thermal expansion of the waters sufficient to drive a transgression of ~10-20 meters even as the sulfate in those brines converts anoxic waters to euxinic ones.  An end-Permian example of the spread of anoxia in deeper waters followed by its upward expansion to shallow settings, and the shift to euxinic conditions is documented in the Sverdrup Basin of Arctic Canada (Grasby et al., 2009) is consistent with the model steps suggested in Kidder and Worsley (2010).

Ocean acidification should develop and intensify as HEATT episodes unfold.  Such acidification has been well documented for the PETM HEATT episode (e.g. Zachos et al., 2005), and Kiessling (2011) recently compiled a list of intervals that bear evidence of having undergone recognizable ocean acidification.  Not surprisingly, all of these are a subset of our list of proposed HEATT episodes.

The coincidence of mass extinctions with HEATT episodes (Table 1) is not trivial.  The combined overall effect of systemically-linked changes in ocean temperature, sea level, anoxia, euxinia, acidification, nutrient crises, and more is likely to be more potent than any of these single factors acting alone.

More on HEATT-Episode Triggers

We suggest that several factors govern the intensity of a HEATT episode.  One is the size of the LIP.  Figure 1 shows approximate minimum areas for LIPs (many of these data are from the Large Igneous Province Commission Website:  http://www.largeigneousprovinces.org/).  It is crudely apparent from that figure that HEATT episodes tend to be associated with the larger LIPs, but since we do not know how much of the original LIP has been destroyed after emplacement, this approximation is just that.  A second important factor is the nature of the crust through which the LIP intrudes.  If the crust is more carbon-rich, the warming potency of a given LIP is greater than a LIP that intrudes carbon-poor crust (e.g. McElwain et al., 1999; McElwain et al., 2005; Erwin, 2006; Retallack and Jahren, 2008; Sobolev et al., 2011).  Host crust composition is also important because if a LIP intrudes through a light carbon source (e.g. coal, black shale), not only is more warming likely, but the negative δ¹³C excursion that marks many HEATT episodes can more readily be explained.  It was shown long ago, for example, that CO₂ emissions alone from Siberian Traps LIP were insufficient to explain the end-Permian δ¹³C anomaly.  A light carbon source (e.g. methane) will solve the problem (see Erwin, 1993; Kidder and Worsley, 2004 for discussion).  Alternative explanations for the Late Permian negative δ¹³C shift (Berner, 1989; Broecker and Peacock, 1999) are elaborated upon in Kidder and Worsley (2004).  Ocean release of methane from warm oceans has separately been considered for the Paleocene-Eocene Thermal Maximum (PETM) rapid global warmup (e.g. Kennett and Stott, 1991; Zachos, et al., 2008).  The North Atlantic Volcanic Province (NAVP) has been suggested as a warming trigger (Svensen et al., 2004; Storey et al., 2007), but it has been emphasized that volcanic CO₂ alone cannot account for the PETM negative δ¹³C  shift (Zeebe et al., 2009).  Recent work (Rampino et al., 2013) suggests that thermogenic methane emissions were significant during NAVP activity.  Third, preconditions that are operating when the LIP erupts probably influence the intensity of the HEATT episode.  If Earth is already in an icehouse or if significant forcing factors are in place to promote planetary cooling, LIP emissions appear insufficient to force hothouse climate.  One example of this is the Middle Miocene Columbia River Basalts LIP.  This LIP erupted during icehouse climate.  Despite its small size, it appears to have forced an increase of global average temperature of ~2-3°C.  The much larger Ethiopian Traps LIP failed to generate warming.  This is probably the result of strong cooling preconditions as Antarctica underwent glaciation at the onset of the Cenozoic icehouse climate.  See further description of these Cenozoic examples and references in Kidder and Worsley (2012).  Additional Eocene-Oligocene factors we did not consider earlier may include limited CO₂ emissions from the Ethiopian traps relative to continental-crust-based LIPs and Cather’s (2009) report of significant silicic LIP activity at this time.  The latter would introduce significant ash deposits that would fertilize photosynthetic activity so as to temporarily offset the warming effects of the Ethiopian traps.  Our sense (Kidder and Worsley, 2012) is that the major climate shift into the Cenozoic icehouse was initially set in motion by the long-term drawdown of atmospheric CO₂ by the Himalayan orogeny, and sharply intensified by the initiation of the Antarctic circum-polar current (e.g. Kennett et al., 1974) that was accompanied by rapidly falling atmospheric CO₂ levels (Pagani et al., 2011) at this time.  We speculate that this fundamental systemic shift dwarfed the influence of both basaltic and silicic LIP activity.

The warming that results from basaltic LIP emissions appears substantial enough to drive the onset of the haline mode, which, in turn, is a warming feedback.  Although precise data are lacking, we speculate that the haline mode can last only as long as the LIP trigger sustains it.  We suggest that a series of LIP eruptions should sustain or revive the hothouse condition. 

We note here that other triggers are possible.  The most likely alternative (or complementary) trigger is bolide impact in carbon-rich rocks.  The Chicxulub impact in carbonate and sulfate sedimentary rocks of what is now the Yucatan Peninsula is one example of potential warming (e.g  Kawaragi et al., 2009).  However, competing short-term cooling influences from aerosol emissions must also be factored in (e.g. Pope et al., 1997; Pierazzo et al., 2003).  Another potential impact trigger is the recently suggested end-Permian impact in and associated with fracturing of Brazilian oil shale beds (Tohver et al., 2013).  Both of these coincide with significant LIP activity.  It is unclear at present whether either geologically instantaneous event would force a hothouse climate on its own, but the light isotopic carbon released by them would help to explain negative carbon isotope anomalies that are much harder to achieve with CO₂ from LIPs.

A challenge to extending the HEATT model into the Early-Middle Paleozoic has been the spotty record of LIP activity.  For example, the Kalkarindji LIP is a tempting trigger for the Botomian (Cambrian) HEATT episode (Fig. 1), but age relations do not yet clearly support a temporal linkage (see Kidder and Worsley, 2010 and related references therein). 

One consideration is that some LIPs that were active in the ancient record may have been overlooked because they left only indirect evidence.  Kidder and Worsley (2010) and references therein suggested that low values of ⁸⁷Sr/⁸⁶Sr hinted at the presence of LIP activity in the Late Devonian.  Although Late Devonian LIP evidence on the East European craton was present, age dates did not clearly link HEATT and LIP activity.  Late Devonian LIP activity has subsequently become better recognized. Recent age dating by Courtillot et al. (2010) and new results reported by Ricci et al. (2013) now temporally link LIP triggers to both the Frasnian-Famennian and Hangenberg HEATT episodes.  Earlier Devonian pulses of warming (e.g. end-Givetian) appear to approach, but perhaps not quite achieve hothouse conditions may be driven by as yet undiscovered triggers.

Support for LIP-induced warming in the Late Ordovician remains elusive, but there is some suggestive evidence.  As with the Late Devonian, low values of ⁸⁷Sr/⁸⁶Sr in the Late Ordovician (referenced in Kidder and Worsley, 2010) may reflect LIP activity.  Lefebvre et al. (2010) suggested a possible Late Ordovician superplume.  We also noted that the ophiolite record may be critical in helping to identify old ocean LIPs that may have been partially obducted as they reached ocean trenches.  Vaughan and Scarrow (2003) have tabulated Phanerozoic obduction frequency, and conspicuous peaks in ophiolite abundance (ghost LIPs?) mark horizons that are associated with some HEATT episode.  There is such an obduction pulse in the Late Ordovician.  We acknowledge Keith Milam for suggesting this possibility to us in 2008.  However, we remain cautious at this point because at least one obduction pulse (Early Permian) has no corresponding HEATT episode.  Icehouse preconditions may explain that lack, but since we have not yet fully explored the obduction pulses as smoking guns for oceanic LIP activity, we refrain from pursuing the matter further at this time.

Still, we predict that LIP triggers are likely for Paleozoic HEATT episodes. Figure 1 illustrates that transgression-associated extinctions are conspicuous over the past ~250 million years (e.g. PETM, end-Cretaceous, Cenomanian-Turonian, Aptian, Toarcian-Pleinsbachian, end-Triassic, end-Permian).  All of these coincide with significant LIP activity.  The end-Jurassic extinction also coincides with transgression (Haq et al., 1987), and the Shatsky Rise LIP (Eldholm and Coffin, 2000; Neal et al., 1997) is emerging as a trigger (Mahoney et al., 2005) for an event that was probably a HEATT or near-HEATT episode.  Figure 1 shows that the Early-Middle Paleozoic is also marked by a number of transgression-associated extinctions (e.g. Botomian, SPICE (see Kidder and Worsley, 2012), Late Ordovician glacially-interrupted warming (see Kidder and Worsley, 2010), Pragian, Givetian, Frasnian-Famenian, and Hangenberg).  A clear record of corresponding LIPs is lacking for most of these Paleozoic examples, but the general similarities between Paleozoic and younger HEATT episodes lead us to suggest that this dearth is probably a function of poor to absent LIP preservation rather than lack of LIP activity.

References 

[1] Present Address:  816 N Marshall St, Boise ID 83706