Australia: The Land Where Time Began

A biography of the Australian continent 

Supercontinent Cycle and Mantle Convection

The distribution through time of large igneous provinces (LIPs) and giant dike swarms (GDS) both are indicated to occur periodically as the result of mantle insulation when supercontinents are completely assembled, according to the authors¹. The authors¹ say there are 7 possible supercontinent events that have been defined by these data over the last 3 Gy, the periodicity of these events being consistent with mantle and supercontinent cycling rates that have been previously estimated. At 1267 Ma the Mackenzie Dike Swarm and at 250 Ma the Siberian Traps have been found to be synchronous with 2 dramatic increases in the GDS rate of production. Since 3.0 Ga 3 episodes of mantle activity have been defined. There is a 475 My gap in the LIP record that occurred at a time when the continents were relatively dispersed, from about 725-250 Ma. The LIP gap is coincident with a period in the history of the Earth that is characterised by marine isotope rations of oxygen, carbon, strontium, and sulphur that are indicative of small mantle fluxes.

It has been suggested by a number of researchers that when a supercontinent assembles the crust shields the underlying mantle resulting in a modification of convective activity, such as mantle plume and positive geoid anomalies, the resulting rifting eventually causing the fragmentation of the supercontinent [1-6]. It is believed that these mantle plumes produce such geological results as large igneous provinces and giant dike swarms [7-8]. There is increasing evidence, according to the authors¹, that these plumes originate near the core-mantle boundary, though there is still debate about the dimensions and generation depths of these plumes [9-11], leading the authors¹ to say that the distributions of the LIPs and GDS should be distributed through time in concert with formation of supercontinents and their timing should be comparable to supercontinent cycles (about 500 Ma; [3-12]). In this paper the authors¹ examine the temporal relationship of LIPs and GDS as proxies for cycles of supercontinents and mantle convection.

According to the authors¹ LIPs are bodies that are comprised of extrusive, typically tholeiitic basalt with associated mafic rock with volumes of more than 100,000 km³ [13]. It is believed that basalt flows produced by these events have resulted from violent eruptions from fissures formed as a result of the interaction of a mantle plume with crust that is either of continental or oceanic origin. An origin near the core-mantle boundary has been suggested by many for these plumes [14-16]. It has been argued by Anderson [17] that mantle plumes may not be required to produce such phenomena. It has been shown by radiometric dating that these eruptions occur over periods of time that are geologically short (105 to 10 6 years [13; 18-21].

Mafic dike swarms have been defined by Ernst et al., 'concentration of dikes of basic composition emplaced in the same igneous episode'. Giant dike swarms are dike swarms with lengths ≥ 300 km [22,23]. Giant radiation dike swarms are those with a radial pattern about a central point [22,23]. The following evidence relating giant radiating swarms to mantle plumes has been cited by Ernst et al., [22].

  1. The central magma source is indicating the radiating pattern of the dike swarm;
  2. a period of time less than a few million years covers the emplacement of the entire swarm;
  3. the presence of volcanic and plutonic rocks at the centre of the swarm that are coeval, so may be the remains of LIPs;
  4. An impinging mantle plume may be indicated by topographic uplift at the centre of a dike swarm;
  5. evidence in the dikes of lateral magma flow that isn't present in the central source area.

The centres of mantle plumes have been defined by Ernst et al. [24] as the focal points of giant radiating dike swarms, the component dikes being of a similar age to each other, and with a radiating pattern that converges on a point that is believed to be the locus of the mantle plume centre. There are 27 possible mantle plume sites identified by the correlation of occurrences of giant radiating dike swarms, based on radiometric age and palaeogeographic reconstruction [24].

Age distributions and LIP preservation, giant dike swarms and plume centres

Covering the period of the last 250 My there are 18 LIPs that have been documented [25], though none have been documented between 723 and 250 Ma. In the Precambrian there are 9 LIPSs that have been documented (723 Ma [26]; 800 Ma [27]; 1108 Ma [28]; 1267 Ma [29]; 2450 Ma [30, 31]; 2705, 2715, 2740 and 2770 Ma [32, 33]. It is suggested by the temporal distribution of LIPs and GDS that there are 7 major episodes of mantle activity over the last 3 Gy, 2800-2700 Ma, 2550-2400 Ma, 2250-2000 Ma, 1900-1600 Ma, 1350-1000 Ma, 850-550 Ma and 350-0 Ma. Since about 1300 Ma there is a strong correlation between episodes of LIP and GDS episodes and plume centres. It is suggested by the authors¹ that this mantle activity is directly related to the assembly of supercontinents.

The authors¹ based the timing of assembly of supercontinents mainly on the observed relationship coupling LIP production and the assembly of Pangaea [34]. Due to the imprecision of various factors used in the timing of supercontinent aggregation and fragmentation the authors¹ assigned secondary importance to various palaeogeographic reconstructions and interpretations [35-37]. Figure 1¹ shows their synthesis of supercontinent cycle reconstructions. They say the timing of the supercontinents that came before Rodinia is relatively unconstrained.

It is shown by the temporal distribution of LIPs and GDS that they occur periodically ([24, 38]; Fig. 1 & 2). Because of the high potential for preservation of the GDS resulting from them being deep intrusive structures that occur vertically in the crust. The result is that the impact of erosion on GDS is relatively minor allowing them to be well preserved, making the GDS record relatively more complete for the the last 3 Gy (Fig.2). Dikes with both map dimensions of length and width that have been reported with the age constrained to better than ± 50 My. The error was more than  ± 50 My in 3 instances, however these ages correspond with other dikes from the same region that have well-constrained ages.

Using regular GDS periods of activity with a constant occurrence rate, punctuated by periods with no activity, the cumulative curve of GDS occurrences can be modeled. Superimposed on this production periodicity a modest loss of occurrences from erosion and tectonic processes (an exponential decay function with a half life of 1386 My( [39]; Fig. 2).

Survival Rate x Periodic Production Rate = Occurrences Remaining after Decay

where Survival Rate = e-kT and the periodic Production Rate = 3.5 occurrences per 50 My for 350 My (supercontinent assembly) punctuated by periods in which there is no production for 150 My (dispersed continents). The periodic production of GDS alone produces 147 occurrences. 78 GDS remain (compared with 79 actual occurrences) when a decay function is superimposed on this function. The number of remaining occurrences are summed and the cumulative % is calculated. This model can also be used for carbonatites and Kimberlites [38, 40].

Following the method of Veizer et al.[38] different values have been calculated for k: f=giant dike swarms (half-life = 1386 My), k - 5.00 x 10-4 ; carbonatites (half-life 450 My), k = 1.54 x 10-3; kimberlites (half-life 200 My), k = 3.46 x 10-3 . A significantly lower preservation potential  than giant dike swarms is indicated by the frequency distributions of carbonatites and kimberlites. The overall of all 3 trends are similar, e.g. the timing of plateaus. The rather simple model used by the authors¹ suggests that giant dike swarms, carbonatites and kimberlites can be produced when using the same periodic driving mechanism in conjunction with a unique preservation potential or half life (Fig. 2b).

According to the authors¹ the excellent fit of their model output suggests GDS occurrence production is periodic, assuming there was some variation between 300 Ma and 500 Ma in supercontinent cyclicity, though the actual timing of the assembly and fragmentation of supercontinents may not be as regular as it is in the model. In the model they held the GDS production rate constant over time. The model deviates from the actual dike occurrence data between about 1.2 Ga and 700 Ma. That the production rate of GDS during that time period may have been higher than modeled is suggested by this deviation. The model overpredicts the number of dike occurrences prior to 2.2 Ga. The authors¹ say the production rate during this time may have been less than predicted by their model. The authors¹ suggest poor preservation of the early GDS record could possibly be the reason for the overestimate by the model. According to the model the rate of survival only affects the absolute number of occurrences at a given time, not affecting the observed periodicity in the accumulative distribution of occurrences of dike swarms. The same temporal distribution observed in all diabase dike swarms is observed in the periodicity of the GDS (Fig. 4; [24]).

The periodicity of the greenstone belt occurrences is grossly similar to that of GDS and the greenbelt occurrences have been associated with the assembly of supercontinents [41]. It has been noted that greenstone belts have not been found between 2.45-2.2 Ga and between 1.65-1.35 Ga (Condie, [41]), roughly corresponding with gaps in occurrences of GDS at 2.4-2.25 Ga and 1.6-1.35 Ga, (Fig. 1). The authors¹ suggest it has not been well established that there is a relationship between greenstone belts and mantle plumes [42], the term 'greenstone belt' being defined broadly and encompassing a wide range of volcanic rocks and sediments associated with them, the authors¹ deciding to avoid study of their temporal distribution in this study.

The compilation of LIPs occurring after 250 Ma (Yale [25]) found less then 1/3 of the LIPs are oceanic. Before 150 Ma oceanic LIPs are not part of the data set as a result of oceanic plate subduction. The authors¹ suggest that there is no reason to think that oceanic LIPs would not be subducted, as oceanic LIPs and oceanic crust have the same composition. It has been found that much of the Ontong Java Plateau is being actively subducted at the present (Taira et al.; [43]). A bias that cannot be corrected for exists towards continental magmatic activity in the LIPs data as a result of continuing subduction. Mantle plume generation associated with pre-150 Ma LIPs data is argued to be related to the assembly of supercontinents and mantle insulation, thus justifying the use of pre-150 Ma LIPs data [1-6]. The authors¹ suggest this bias doesn't preclude examination as it is the supercontinent record that is being searched for.

Periodicity of processes controlling LIPs eruptions is suggested by the temporal distribution of LIPs and intervening gaps (Fig. 1). The authors¹ suggest it has not been possible to accurately calculate the exponential decay function because of the small number of LIP occurrences (N = 27) [39]. The large gaps in the frequency of LIPs distribution cannot be entirely explained by lack of preservation, assuming a constant rate of production of LIPS and the use of an exponential decay function that accounts for the loss of LIPs occurrences over time.

The preservation of LIPs erupted over the last 1 billion years appears to be relatively good, as suggested by the occurrence of several LIPs between 3 and 1 Ga. The conclusion reached is that there is inherent periodicity in the LIP record, though diminished preservation with age modifies it to some degree. Combining the LIP, GDS and plume centre data is believed to accurately represent the mantle plume temporal distribution, and thus the vigor of mantle convection. Supercontinent assembly and fragmentation is the causal mechanism for this periodicity [1-6].

Pangaea and LIPs

The assembly of Pangaea and the correlated cumulative volume of LIPs erupted after 250 Ma is the best example of the correlation between supercontinent assembly and LIP eruptions. The rate of change of volume of LIPs extruded increases modestly from 250-200 Ma, and rapidly from 200-60 Ma, plateauing over the last 60 My, following the assembly of Pangaea (Fig. 3a). About 70 Ma after Pangaea assembled the initiation of LIPs eruptions is consistent with Anderson's [1] estimate of about 100 My for the formation of a 50 m geoid anomaly under a supercontinent (Fig. 3a). It has been suggested that the ascension rates of plumes could be as high as 30 cm/yr (Courtillot & Besse, [45]). At 5 cm/yr [40], a more conservative rate, and a distance travelled of about 2,900 km (depth D") yields a time of 58 My (Fig. 3a). These estimates are consistent with the initial assembly of Pangaea at 320 Ma and the timing of initial eruptions of flood basalts at 250 Ma.

Between 150 and 70 Ma the volume of LIP maximum increase reached a peak, the rapid volume increase period coinciding with the superplume event of the mid-Cretaceous between 122 Ma and 83 Ma that has been described by Larson [15] and Larson and Kincaid [16]. During this time period there are extrusive LIP volume data indicating elevated volume relative to a 4th-order polynomial fit through the other data, the deviation being suggested by the authors¹ to probably be the result of mantle fluxes that were elevated during this interval ([25]; Fig.3a). The LIP production rate that declined over the last 60 My suggests the effects of mantle heating that had been induced by the formation of Pangaea was waning at that time, the decline corresponding with Alpine orogenic activity and the uplift of the Himalayas. The marine strontium isotope record over the last 50 My may be at least partially explained by these mantle fluxes that are potentially lower.

Supercontinents before Pangaea

Mantle activity episodes

According to the authors¹ there were 3 distinct periods of GDS activity on a plot produced of cumulative map area of GDS versus age over the last 3.0 Ga (Ernst et al., 1996):

Period I     2771-1300 Ma

Period II    1270-250 Ma;

Period III    250-0 Ma:

(Fig.5). Each of these periods is defined by a linear relation that is highly significant, r² = 0.94, 0.98 and 0.90 respectively. At 2 important points in the history of the Earth slopes of 2 of these trends increase dramatically; Mackenzie Dike Swarm emplacement, 1267 Ma, the eruption that produced the Siberian Traps, 250 Ma (fig.5). Significant increases in mantle-derived magma production, 56 % and 61 % respectively, are represented by the trends in Fig.5 if the extent of GDS area is closely related to the volume of flood basalt and the GDS record is well preserved (Fig.2). The authors¹ say it is no coincident that the dramatic slope changes in these tends occur in synchrony with 2 of the largest igneous events in the entire history of the Earth, and that these events occur as supercontinents are assembling, Rodinia and Pangaea (Fig.5). The authors suggest the assembly of supercontinents is the causal mechanism for these changes, mantle insulation and mantle convection modification. The authors¹ suggest this statement implies that the need for an increase of size over the last 3 Ga (areal extent and/or thickness), and they also suggest that a possible cause may be the need for a size increase of the continental crust since the Archaean [55].

It has been suggested that within the last 1 Ga there has been a significant modification of mantle convection (Allegre, [9]. As a result of this change the 2 layer convection broke down, allowing the descending slabs, that were cold, to reach the lower mantle, and hotspots/plumes that were generated at the boundary of the core and the mantle to rise to the surface as flood basalts. The authors¹ say this hypothesis is not supported by the data in this [their paper], saying a more plausible model would be that described by Tackley et al. [56], with the 2 mantle catastrophes being associated with Mackenzie Dike Swarm and the Siberian Flood Basalt events (Fig.5). The authors¹ say that periodically since the close of the Archaean LIPs/flood basalts have erupted, that they say suggests several perturbations of the a 2-layer mantle convection scheme have occurred over the last 3 Ga.

The Early Palaeozoic LIP gap and changes of ocean chemistry

After the assembly of Pangaea the LIP volume increased (Fig.3a), that authors¹ saying that this strongly suggests that mantle fluxes to the ocean should be positively correlated, as there is an interaction between plumes and continental and oceanic crust (e.g., LIPs and hydrothermal activity). Mantle flux increases would lower seawater 87Sr/86Sr ratios and δ 34S values and increase marine δ 18O and δ 13C values [25, 57-59]. These isotopic changes should in general mimic supercontinent cycles, though with lag times that are appropriate for different elements (Fig.6).

The LIP gap, a lack of LIPs, occurs in the period of time between the initial assembly of Greater Gondwana and Pangaea, 723-250 Ma (Fig.6). This LIP gap coincides with the time period covered by the Late Neoproterozoic and the Early Palaeozoic, a time that was characterised by marine carbonates with low δ 18O (and δ 13C values)([58]; Fig.6). It has been suggested the the low δ18O that have been found for seawater in the Early Palaeozoic, from the Cambrian to the Devonian, resulted from the dominance of low temperature silicate weathering reactions versus exchange reactions that were associated with hydrothermal seafloor activity at high temperatures (Walker & Lohmann, [57]; Carpenter et al., [58]) [60]. The 87Sr/86Sr ratios [61] and δ 34S values [62] of this period of time support this hypothesis and suggest continental fluxes were dominant over mantle fluxes [59]. During the Late Neoproterozoic and Early Palaeozoic the lack of LIPs adds to the evidence that mantle fluxes were lower in this time interval (Fig.6).

 The Blue Mountains of the eastern US metabasalts from the Catoctin Volcanic Province have been dated to 570 ± 36 Ma [63] and metarhyolites 572 ± 5 Ma [64]. This occurrence doesn't meet the criteria for size of 100,000 km³of Coffin & Eldholm [13] so was excluded from the study by the the authors¹, and the occurrence of metasediments and metarhyolites intercolated with the Catoctin metabasalts is not characteristic of other LIPs that have been reported [63-65]. The areal extent of the Catoctin metabasalt [63, 64, 66, 67] is smaller those reported for LIPs that were used in this study by several orders of magnitude [13, 25], though the Catoctin metabasalt is reported to be 2 km thick [64] that is typical for other reported LIPs. The LIP gap would span an interval of 320 My from 570-250 Ma, instead of from 723-250 Ma (Figs.1 &b 6), if this basalt is considered to be a LIP.

It has also been recognised that about 350 Ma intracratonic basins began to rapidly subside (Kiminz & Bond, [68]), indicating the initial aggregation of Pangaea above a downwelling, cold region. The formation of a plume began once Pangaea had assembled and began to thermal blanket the mantle beneath, the plume eventually culminating in the Siberian flood basalt and the fragmentation of Pangaea. The authors¹ suggest the subsidence in the Middle Palaeozoic marks the transition from a number of dispersed continents and a cool mantle to an assembled Pangaea and a warm mantle.

As oxygen remains in seawater for a long time (about 40-100 My, [69, 70]) there is a perfect correlation between marine values of δ18O, and LIP occurrences are not expected. At the Devonian-Carboniferous boundary, about 350 Ma, [58, 71], there was a shift to a higher δ18O value that was globally significant. The about 4 ‰ change occurred over about 10-20 My [71], on the basis of empirical measurements of marine carbonates. According to the authors¹ correlation of this increase in δ18O values with the beginning of the assembly of Pangaea suggests supercontinent assembly and mantle insulation is the mechanism for a change from low temperature to high temperature dominated silicate exchange reactions. The authors¹ consider it unlikely that mantle heating from the previous formation of a supercontinent induced isotopic changes at the boundary of the Devonian and Carboniferous, as the previous supercontinent assembled in the Late Neoproterozoic. They suggest the beginning of a period of mantle quiescence, coinciding with the dispersal of the continents, may be reflected by the absence of LIPs in the Early Palaeozoic, the mantle flux reduction persisting until Pangaean assembly began (Fig.6). This is believed to indicate that Gondwana, that was present during the Early Palaeozoic, was too small to alter mantle convection.

At the end of the Proterozoic Eon dramatic evolutionary changes coincide with geochemical changes, possibly resulting from an abrupt decrease in the activity of hydrothermal vents on the ocean floor and an increase of dissolved O2 in seawater [59, 72-74]. It has been argued that there is period  when seafloor hydrothermal activity was reduced at the end of the Neoproterozoic and Early Palaeozoic, based on marine carbon and sulphur isotope data (Carpenter & Lohmann, [59]. The authors¹ suggest the increase in O2 that was necessary for the dramatic evolutionary changes that have been recorded during the Vendian (Ediacaran) [59, 72] may have resulted from the reduction of ocean floor hydrothermal activity. In the Late Permian the onset of LIP eruptions, and the seafloor hydrothermal activity that was associated with it, may have had an impact on life that was equal but opposite [75]. Since 800 Ma the LIPs record is consistent with mantle flux reduction near the Proterozoic-Palaeozoic boundary.


New proxies for the assembly of supercontinents and mantle convection, that are consistent with marine isotopic records, are giant dike swarms and large igneous provinces. The autors¹ suggest to better constrain the age and origin of the potentially useful proxies, GDS and LIPs further geochemical analyses of them is required. They say the events associated with the Mackenzie Dike Swarm and Siberian Traps are particularly significant and need further study.

See Source 1 for the references mentioned in the text.

Sources & Further reading

  1. Large igneous provinces and dike swarms : proxies for supercontinent cyclicity and mantle convection




Author: M. H. Monroe
Last Updated 02/06/2012

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