Australia: The Land Where Time Began

A biography of the Australian continent 

Severe Selenium depletion in the Oceans of the Phanerozoic as a Factor in 3 Global Mass Extinction Events

The key trace element that is required by all animal and most plant life is selenium (Se) and deficiencies of Se in the food chain lead to pathologies or death. In this study Long et al., show that, based on new geochemical analyses of trace elements in marine pyrite from the Phanerozoic that sustained periods of severe Se depletion in the oceans of the past correlate closely with 3 major mass extinction events, at the end of the Ordovician, Devonian and Triassic periods. Periods of Se depletion >1.5-2 orders of magnitude lower than abundance in the oceans of the present, being within the range where it could cause severe pathological damage in extant organisms that are Se-reliant. Long et al., suggest that selenium may have been one of several factors in these complex extinction events. It is considered likely that recovery from the depletion/extinction events is part of a natural marine cycle, though rapid rises of global oxygen that result from sudden major increases in marine productivity and biomass of plants following each extinction event may also have played a crucial role.

The formation and sustainability of life many trace elements (TEs) require many essential trace elements (Mertz, 1957; Klasing, 1998; Eisler, 2000). In this study Long et al. used as a new dataset of TE abundances in oceans of the past (Large et al., 2014; Large et al., 2015a) in order to discuss whether 3 of the 5 mass extinction events that occurred in the Phanerozoic could have been influenced by extreme low abundances of selenium, a trace element.

Such TE have been measured by the use of new laser ablation-inductively coupled plasma mass spectrometry (LA-ICPMS) techniques with accuracy down to single ppb in pyrite from marine black shales. Essential TEs in the ocean fell below critical thresholds during 3 mass extinction events at the end of the Ordovician, Devonian and Triassic, as has been revealed by  this dataset of TE throughout the past 3 Gyr (billion years), initially based on some pyrite analyses from 1885 (Large et al., 2014), and now updated to include more than 2,200 analyses for the Late Neoproterozoic-Phanerozoic (Large et al., 2015a). The mechanisms for these extinction events are debated, though global climate change that was associated with widespread anoxia and changes of eustatic sea level (Ordovician, Devonian) as well as the Central Atlantic Magmatic Province (CAMP) eruptions (Triassic) were involved. In this paper Long et al. present evidence that is based on known environmental and tolerance levels of selenium in a range of extant organisms from phytoplankton to vertebrates to propose the way in which distinct periods of Se depletion in the oceans of the past offer a potential new causal factor in these mass extinction events.

Selenium, weathering and oceans

Selenium is a metalloid that occurs naturally, which is unique in that it may be toxic at high concentrations, though it is an essential micronutrient in most organisms, including bacteria, archaea, fish and shellfish, and has a concentration window for sustaining life in the oceans of the world, as is discussed further below. In the crust of the Earth the content of selenium averages 0.05 ppm, with the highest levels being found in shales (up to 675 ppm), coals (up to 20 ppm) and volcanic tuffs (up to 9.2 ppm), compared with igneous rocks that range from 0.01 to <2 ppm (Plant et al., 2005). It has also been reported that organic-bearing chalks from the Cretaceous contain up to 70 ppm Se (Kulp & Pratt, 2004).

There are multiple oxidation states of Se -2, 0, +4, and +6.

1.    under oxic conditions Selenate (Se6+) is the predominate inorganic species;

2.    Under suboxic conditions selenite (Se4+) is predominant, but

3.    Under anoxic to euxinic conditions selenide (Se2-) organo-selenium complexes and elemental Se0 are most stable.

Selenite and selenate are highly soluble, though selenite is readily absorbed onto iron oxides and organic matter, especially in environments where the pH is low, which leads to its retention in the soil profile under favourable conditions (Neal et al., 1987). Consequently, selenite is less bioavailable than selenate (Fordyce, 2007). Oxidation of Se4+ to Se6+ enhances the mobility of Se and persistence in natural waters.

Disseminated pyrite in sedimentary and volcanic rocks where the Se substitutes for sulphur (S) in the structure of pyrite is the principal source of weathering of the crust. From <0.5 to 5209 ppm Selenium with an arithmetic mean of 145 ppm is contained in sedimentary pyrite (Large et al., 2015a). The release of selenium as both selenate and selenite species results from oxidative weathering of pyrite. The selenate species remains highly soluble under neutral to alkaline conditions, where it can readily be transported by river systems to the ocean (Cutter, 1989). Continental weathering releases little soluble Se as selenide under reduced conditions, and elemental Se (Seo) and organo-selenium complexes are relatively insoluble (Kulp & Pratt, 2004). A major increase in the soluble Se to the ocean therefore results from significant increases in atmospheric oxygen, accompanied by active erosion. The degree of oxidation and pH controls the ratio of selenate to selenite of the selenium that is released, which in turn controls the solubility of Se in surface runoff, and consequently the amount of Se that is released to the oceans, if other factors remain constant. It is indicated by studies of dissolved TE supply from rivers to the ocean (Kharkar et al., 1968) that Se has a narrow range of concentration, similar to Ag, Co, Rb and Cs, and in contrast to Mo, Cr, Sb which exhibit extremely large variations from river to river. It was also shown by this study that most of the stream load of Se was dissolved, and only 10% adsorbed to the particles, and when it made contact with the seawater it was desorbed. There are 3 species that are present in the marine environment, selenite, selenate and a form of organic selenides (Cutter & Bruland, 1984). Decomposition of selenite-silicate compounds provides the selenite in the hard tissue of phytoplankton in the tropical and subtropical parts of the Pacific Ocean (Nakaguchi et al., 2008). Surface waters down to about 200 m are dominated by organic selenides (seleno-amino acids in complex peptides), below which selenite and selenate are the major dissolved species (Cutter & Cutter, 1995). In the upper layers of the ocean the low concentrations of dissolved oxidised Se results from biochemical reduction of oxidised species into labile organic particles during assimilation by marine organisms, which sink, die and dissolve, and consequently regenerating the dissolved selenate and selenite species at lower levels in the ocean, which is typical of a nutrient profile. The oxidised Se species are reduced to Seo         or Se- in the deep ocean, and are incorporated into pyrite (Mitchell et al., 2012). A small fraction of the selenium that is organically fixed eventually deposits in muds on the seafloor (Herring, 1991). It was shown (Ryser et al., 2005) that selenium in black shale is present as products of anaerobic microbial respiration that result from the microbial reduction of Se oxyanions, which includes Se-substitutes for sulphur in pyrite, di-selenide carbon compounds and dzharkenite (FeSe2: an isometric polymorph of ferroselite). It was indicated by experimental studies (Diener et al., 2012) that Se2- is taken up (98%) by pyrite to produce a FeSSe compound with a pyrite structure that is distorted slightly Ferroselite. (FeSe) has also been produced by reacting nanoparticles of pyrite and greigite with selenite and selenate solutions (Charlet et al., 2012). It was indicated by the LA-ICPMS imaging of pyrite in black shales of various metamorphic grades that pyrite is enriched considerably in Se compared with the clay-rich and organic-rich matrices. The analytical data of Long et al. indicate that in selenium in pyrite is enriched, 5.8 times on average, over the selenium content in the black shale matrix (which commonly contains microscopic pyrite grains). This compares with Mo which is enriched only 2.5 times relative to the matrix.

Se depletion and mass extinctions

Throughout the Phanerozoic the overall trend for levels of Se that were observed showed that selenium dropped dramatically relative to modern levels in the oceans below critical thresholds during 3 mass extinction events at the end of the Ordovician, Devonian and Triassic. These 3 key biotic crises are analysed further and discussed in detail below.

End-Ordovician extinction event

Commencing about 455 Ma, the end-Ordovician extinction was a complex event that ended with 2 pulses, 445 Ma and 443 Ma, which were preceded by glaciations of the South Pole (Harper et al., 2013). Long et al. suggest it is possible that dramatic changes in sea level coupled with cooling of the tropical ocean played a role in the extinctions (Finnegan et al., 2012), while it was suggested by others that euxinia and a sudden drop in oxygen caused the first pulse, and transgression of anoxic water onto continental shelves drove the second pulse (Hammerlund et al., 2012). A massive drop in Se levels in pyrite is shown by the Se curve from a peak at 523 Ma of 548 ppm Se (geometric mean) or 365 ppt for seawater Se by use of the concentration factor that was outlined above, to lows of about 2 ppm (Se in pyrite) equivalent to about 1 ppt (seawater Se) by 455 Ma, which equates to approximately <1% of the current levels in the oceans of the present, well within the Se deficient zone based on known tolerance levels for many extant marine organisms as discussed above. Levels of Se then rise steeply into the Silurian.

Following 2 prolonged pulses of Se depletion the 2 end-Ordovician extinction events occurred. It has been documented that the first of the major extinctions occurred in the latter half of the second phase of Se depletion (~457-449 Ma), with select families of brachiopods (Foliomeria and Probosciambon faunas, early virgianid faunas which declined from about 450-448 Ma (Sutcliffe et al., 2001). In the oceans extreme Se deficiency fluctuated for about 13 Myr before the first major pulse of extinction took effect, and the second, most severe pulse at the end of the Hirnantian only 2 Myr later (Harper et al., 2013). At the end of Se phase the extended phase of severe marine Se depletion would have made it difficult for complex organisms that had selenoproteomes to survive, and could therefore affected much of the food chain, and the first species to go extinct would have been specie that were more dependent on selenium. Throughout the Ordovician, and peaking earlier than the Katian (Servais et al., 2010), the great Ordovician biodiversification event took place throughout the Ordovician and was apparently not affected by the Se depletion events at the close of the period.

Middle-Late extinction events

There were 2 pairs of pulses of extinction events that comprised the Middle-Late Devonian extinctions. Beginning with:

1)   Taghanic and Frasne Crises (House, 2002; McGhee et al., 2013; McGhee, 2014), which were followed by 2 events near the Frasnian-Famennian boundary (Kellwasser event; Gereke and Schindler, 2012) and

2)   at the end of the Devonian (Hangenberg event; Sallan & Coates, 2010).

3)   The Mid-Givetian Taghanic event (~385 Ma), is ranked as the 7th most severe biotic crisis in the Phanerozoic, in which 71 families of marine invertebrates went extinct (McGhee et al., 2013).

4)   The Frasnian-Famennian Kellwasser event dated to 373-374 Ma, involved widespread loss of marine species (13-40% loss at family level, 50-60% of all genera, 72-80% loss of all marine species lost; McGhee, 2014).

The Hangenberg events (~359 Ma) in which there were further extinctions with about 50% of all vertebrate diversity lost (Sallan & Coates, 2010). The Kellwasser event has been characterised by the spread of oceanic anoxia (Riquier et al., 2006), though it may have been restricted to epicontinental shelf seas and not necessarily as widespread as has previously been thought to be (George et al., 2014).

A trend is shown of sequential deficiency at staggered periods between 400 and 350 Ma, by analyses of the bioessential abundance of TE (such as Co, Cu, Ni, Mn, Zn Mo and Cd) through the Devonian. The sequence of deficiency is:

·       Emsian Co,

·       Givetian Mn, Cu, Ni, Zn,

·       Famennian Mo and Se,

·       Famennian-Tournasian Cd.

According to Long et al. the sequential deficiency is most likely to be redox potential and residence times of the respective TE. Under oxidised conditions Co and Mn are least soluble with short residence times, whereas the most soluble with longer residence times are Se, Mo and Cd (Large et al., 2015a). Cu, Ni and Zn have intermediate residence times. As global anoxia increased through the Middle to Late Devonian, therefore, the TE was drawn down sequentially. The peak Mo, Se and Cd deficiencies in seawater over the Famennian to Tournasian period could have affected the marine organisms that are not affected by the sequential deficiencies of Co, Cu, Ni or Zn in the Emsian to Givetian. The Cd, Ni and Zn cycling in the oceans is tied intimately to biogenic cycles (Armour et al., 1985), and Cu is associated with micronutrient cycles and a deep water scavenging process (Daniellson et al., 1985). Co and Cd can both substitute for Zn in diatoms in waters that are Zn depleted so it has been suggested that substitutions by other trace metals or metalloenzymes under certain TE impoverished conditions could be a common strategy for phytoplankton survival (Price & Morel, 1990). Based on the data of Long et al. at certain critical times this strategy may not have been possible, which shows overlapping periods of peak depletion for many elements.

A sudden drop at the end of the Emsian, about 393 Ma, is another observation that arose from the Devonian Se chart. According to Long et al. this corresponds with another series of minor extinction events that occurred in the marine realm, during which many vertebrate groups (most families of osteostracans, galeaspids, heterostracans, and several placoderm families), went extinct (Long, 1993; Janvier, 1996) and these were followed by 2 pulses of extinction in which guilds of invertebrates went extinct in the early and end Eifelian (lower & upper Kacak events; McGhee et al., 2013). 

Just following the Frasnian-Famennian event the longest period of Se deficiency occurred at about 367 Ma. Fish began breathing air at this time leading up to 370 Ma (Clack, 2007; Clement & Long, 2010) and tetrapods, which were the first vertebrates that were equipped to leave the water and venture out on land, also appeared (Clack, 2014). A factor in the attempts by tetrapods to leave the water and move out onto land was possibly related to the collapse of the food chain that resulted from the biotic crises at this time, though it wasn’t until the Early Carboniferous that the complete tetrapod terrestriality was effectively achieved (Long & Gordon, 2004).

SeO32- and SeO2- were likely to have been drawn down as they are relatively soluble under an oxygenated atmosphere as insoluble HSe- species over the long span of the Frasnian-Famennian anoxia. Large inputs of oxygen from the rapid increase in terrestrial biomass over the Middle-Late Devonian could have reversed this pattern (Algeo et al., 2001; Gibling & Davies, 2012). The evolution of plant secondary growth led to an increase in heights of about 2 m in the Middle Devonian to large trees up to 20 m by the Late Famennian (Algeo et al., 2001). Large plant cover increased by from about 10% to 30% at this time and spread from the lowlands and eventually included upland habitats (Gibling & Davies, 2010). The significant increase in global biomass was helped by both these factors between the Mid-Late Devonian, and therefore, a large source of new atmospheric oxygen. It was also noted by Long et al. that massive exhalation of seafloor from mid-ocean ridges and continental margins that was occurring as part of the Variscan orogeny from 356 to 345 Ma accounts for further increase of nutrients into the oceans and thereby bringing about an increase in marine photosynthesis and therefore more oxygen (Tornos, 2006).

End-Triassic extinction event

At the end of the Triassic about 201 Ma extinctions include many major losses in marine as well as terrestrial habitats and are ranked as the second most severe biodiversity crisis in the Phanerozoic. About 20% of families and up to 50% of genera went extinct in the marine realm, including the iconic conodonts (Onoue et al., 2012). Included among terrestrial vertebrate extinctions are large archosaurs, with the exception of the dinosaurs, which paved the way for occupation of niches by radiation of Dinosaurs in the Jurassic.

The Central Atlantic Magmatic Province underwent massive volcanic eruptions that caused a rapid rise in atmospheric CO2 and methane, and led to acidification and localised anoxia that drove marine extinctions, Based on the recognition of shocked quartz impacts of asteroids at this time have been invoked by some authors (Bice et al., 1992), though this has been dismissed as having extinction events that were localised (Onoue et al., 2012). A strong warming effect on land that resulted from greenhouse gases has been suggested by data from long chain n-alkanes that were preserved in fossil plants (Ruhl et al., 2011).

It is implied by the data of Long et al. that extremely low Se levels in the ocean between 202 and 190 Ma that were close to the levels at the end of the Ordovician and Late Devonian as pyrite Se levels are about 2 orders of magnitude lower than pyrite in the modern ocean, and therefore a similar extinction could have been operating as during these earlier mass extinction events. At 202-190 Ma extremely low levels of Cd are lower than at any other time in the Phanerozoic.

The end Triassic, as mentioned above, was a time of in increasing CO2 levels (Royer, 2006) and high fire regimes that were widespread (Belcher et al., 2010). It is shown by one model that O2 levels were rising from about 10% PAL at 205 Ma about 17% Ma PAL by 190 Ma (Falkowski et al., 2005), whereas O2 levels are shown by another dropping from high levels at the end of the Triassic (23% PAL; about 220 Ma) to a peak low at about 180 Ma of 14% PAL, which was followed by a gradual rise (Berner, 2009). The rapid decline in marine Se levels at this time is clear, and might therefore been a more significant factor than in marine extinctions, in spite of conflicting estimates of O2 levels at this time. It has been proposed that gaseous exchange of Se from phytoplankton to the atmosphere was a way in which the biogenic cycle of Se in the ocean can influence directly terrestrial Se levels (Armouroux et al., 2001), so could have affected the terrestrial food chain.


It has not yet been determined what caused the Se and other TE depletions in the ocean, though theoretically it is explainable (Large et al., 2014, 2015,a,b). Increased biogenic productivity in the ocean is driven by increased nutrients and, as a consequence, there is more burial of organic matter which fuels increased oxygen production further as a positive feedback loop (Large et al., 2015a). Se levels increased in the oceans as this regime continued. Subsequently lower amounts of TEs was transported to the oceans as lower atmospheric oxygen or larger areas of the land was covered by higher sea levels, and this resulted in less erosion resulting from less terrestrial oxidation. Rapid drawdown of Se and certain other TEs was caused by this and this led to levels falling below critical thresholds that were necessary for sustaining most marine life. The periods of minimum marine Se may have lasted for up to 10 Myr before rebound, based on the TE data. As a result of lower productivity of the oceans during Se minima, there would have been a dramatic slowdown of TE drawdown; nutrient supply that was related to continental erosion would, however, be on-going. The result of this was a gradual buildup of TEs to the ocean, which activated a rebound of marine life, and a consequent positive feedback of increased atmospheric oxygen. The upturn in recovery of oxygen and Se and recovery from extinction conditions may have, therefore, been a part of the ocean-atmosphere cycles.

Alternatively, jumps in global oxygen from sudden major increases in plant biomass may have broken the Se depletion cycles. The primary invasion of land plants in the Silurian was widespread (Gibling & Davies, 2010). There was a massive increase in the size of land plants in the Late Devonian from predominantly mid-sized forms to (about 2-3 m) in the Middle Devonian, to an abundance of forms of tree size up to 20 m by the end Devonian, and plants occupied much larger areas of the land (Algeo et al., 2001). Major increases in global CO2 that was associated with high floristic turnover also characterised the end Triassic (Belcher et al., 2010). In each of the 3 extinction scenarios it was possible, therefore, that the increase in atmospheric oxygen that was supplied by increasing biomass could have been the significant factor in breaking the anoxia-Se cycle and restoring balance to the marine TE cycle.

According to Long et al. the new data also fit well with the current explanation for the end Permian mass extinction event that was caused by massive volcanic eruptions, in which major marine extinctions are not related to anomalous TE levels (Knoll et al., 2007). Prior to the K-Pg Boundary extinctions at 66 Ma, there are also several Se depletion events. Long et al. found no evidence in the fossil record to suggest that these depletions had any major effect on marine ecosystems at the time, though they noted that the combination of low Sea levels with low oxygen (Falkowski et al., 2005) was only attained during the other 3 prior major extinction events that have been discussed in this paper. At about 93 Ma the global extinction of ichthyosaurs (Bardet, 1992) could potentially reflect the beginning of changes to the food chain that was caused by such effects or might simply be an artefact of poor sampling.


It is suggested by Long et al. that depletion of essential TEs (Se in particular) to levels that were potentially lethal was shown to be correlated highly, and a likely contributing factor, at least 3 mass extinction events in the marine realm, in association with a range of other environmental factors such as global oxygen and carbon dioxide levels that are increasing/decreasing, euxinia and major eustatic sea level changes in the oceans. Oxidative erosion was increased by increased atmospheric oxidation, which released more Se, Mo, Ni, and other TEs into the oceans. Increased biogenic productivity, and consequently more burial of organic matter, which further fuelled increased production of oxygen as a possible feedback loop, was driven by increased nutrients in the ocean. Se levels increased in the ocean as this regime continued. Rapid drawdown of Se and certain other TEs resulted from increased global anoxia in the oceans. Less erosion due to less terrestrial oxidation resulted from lower atmospheric oxygen or higher sea level coverage of land area, therefore lower amounts of TEs flowed back into the oceans, which led to levels falling below critical thresholds that were necessary for a high percentage of marine life. Sudden changes in oxygen levels broke the cycle, possibly as a result of rapid recovery by increased biomass of plants on land responding to high CO2 concentrations, or by invading more land area, as was the case at each of the 3 mass extinction events.

The new data also fit well with the current explanation for the mass extinction event at the end Permian, when major marine extinctions are not seen to have been related to anomalous TE levels. Long et al. do not have Se data across the Cretaceous-Palaeogene boundary, so cannot make any comment in the paper.

The hypothesis of Long et al. is based on a 3.5 Gyr history of trace element abundance in the oceans, though it now needs further refinement from additional data, not only from temporal gaps in their database being filled from pyrite samples, but also on minimal Se and other TE requirements across a wider range of living organisms in order to develop the test models of ecosystem collapse under times of severe TE depletion in the oceans.


Long, J. A., et al. (2016). "Severe selenium depletion in the Phanerozoic oceans as a factor in three global mass extinction events." Gondwana Research 36: 209-218.



Author: M. H. Monroe
Last Updated 09/09/2020 
Journey Back Through Time
Experience Australia
Aboriginal Australia
National Parks
Photo Galleries
Site Map
                                                                                           Author: M.H.Monroe  Email:     Sources & Further reading