David Bercovici, Department of Geology & Geophysics, Yale University

Impatient person’s summary: To understand how plate tectonics began we have to first figure out why we have plate tectonics at all, and how to identify a plate tectonic planet if it were passing by. The uniqueness of plate tectonic is mostly tied up in its odd boundaries, namely one-sided subduction, narrow passive ridges and strike-slip margins that appear to serve no purpose. A lithosphere that forms these weird boundaries requires some exotic physics, mostly involving self-softening, in which the more it’s deformed the softer it gets; and this leads to localized deformation at plate boundaries. The necessary physics to get this behavior is an important goal, and I argue that the pervasiveness of mylonites at plate boundaries is an important clue, in that mineral grain-reduction and resultant weakening, or grain-damage, gives the requisite physics. This grain-scale physics then also provides a framework to infer how plate tectonics started, especially the notion that plate boundaries accumulated over a 1Gy time lag between late Hadean proto-subduction and global plate tectonics by about 3Ga. In particular, early intermittent proto-subduction caused nascent weak damage zones, which persisted as scars even after being abandoned by their pseudo-slabs. These scars were then inherited and rejuvenated as different plate boundaries, after new subduction zones formed elsewhere; eventually, the damage zones gathered into a network of global tectonic plates. This model also allows some framework for understanding the full coupling between plate tectonics and climate, and perhaps even the chicken-or-egg question of what came first, water or plate tectonics.

Complete Blog: This is the third of four blogs written in advance of a meeting at the ETH conference facility in Ascona to discuss four related questions about the evolution of the solid Earth: “When did Plate Tectonics begin on Earth?”, “How did Plate Tectonics begin?”, “What was Earth’s tectonic style before Plate Tectonics began?”, and “Why is it important to understand when and how Plate Tectonics began on Earth, and What came before?”  In this blog, the focus is on the second question: “How did Plate Tectonics begin?”  The March blog “When did Plate Tectonics begin on Earth?” can be found here. The April blog ““What was Earth’s tectonic style before Plate Tectonics began?” can be found here.


Earth is unique amongst the known terrestrial planets for many reasons, with the existence of life and liquid water, a large satellite (our Moon), a strong magnetic field, and of course last but not least, plate tectonics. Plate tectonics probably plays a key role in some of these other features. Most importantly for we organisms, plate tectonics drives the long term geological carbon cycle that stabilizes climate for millions of years [e.g., Walker et al., 1981], and thus probably allowed the right conditions for maintaining liquid water and the development of complex life [see Ward and Brownlee, 2000]. But, answering the question of “how did plate tectonics begin?” assumes we know the answer to why we have plate tectonics now, which would be, euphemistically, a stretch. Ideally we would like to know why Earth has plate tectonics at all, and then we can use that knowledge (or as much as we can get of it) and apply it to understanding how and why plate tectonics initiated. At least that’s the plan (or pipe-dream).

Would you recognize a plate-like planet in a line-up?

If we want to understand why Earth’s plate tectonicness (our blog leader has lauded the free form of our blogs, so I’m taking the liberty of making up words) is unique, we have to have a clue about some identifying features, the blemishes and scars, and the ill-advised tattoos obtained after a night out with the other planets (we all know how hot Venus is). How would you recognize a plate-like planet if you saw one? This of course is a nagging question also in exoplanet astronomy (because of the habitability and climate issues mentioned briefly above); what would a plate-like planet look like from the sparse astronomical observations one can make of very distant terrestrial planets? What would Earth look like if you could only see it from another solar system?

Alfred Wegener’s brilliant puzzle reconstructions aside, the modern theory of plate tectonics (which is not the same as continental drift, since that theory incorrectly would have continents plowing through ocean crust like an ice-breaking ship, which is impossible) arose because of observations of the seafloor. Seafloor topography was mapped out (a lot because of naval operations during World War II – indeed where one of the fathers of plate tectonics, Harry Hess played an important role) and revealed trenches and ridges. And lineations in sea-floor magnetized minerals cracked the case for seafloor spreading (i.e.,the Vine-Matthews-Morley hypothesis). The point is it would be quite difficult to do all that with a telescope and hard enough with dedicated satellites, especially if there is an ocean in the way of what you’re trying to map.

The other terrestrial planets in our neighborhood have no oceans thus satellite observations do tell us enough to say something useful. Mars and Venus in particular have many features of geological activity, and perhaps did have plate tectonics at one time in the past. Mars has perhaps some evidence of seafloor spreading in its distant past, gleaned from magnetic lineations, but probably none now [Connerney et al., 1999, 2005, Yin, 2012]. Venus has signs of resurfacing events, possibly with recurrence intervals of 500 million years [Turcotte, 1993, Turcotte et al., 1999], and perhaps features that look a bit like ocean trenches [Schubert and Sandwell, 1995]. But neither planet has signs of a plate tectonic network of ridges, strike-slip faults and trenches, i.e. the sign of plates being created, sliding along and being destroyed in a continuous cycle of resurfacing. However, if these planets had plate tectonics at one time and it died off, then understanding why would be incredibly important for knowing why Earth still has plates. Nature’s failed experiments are as informative as her successful ones.

But if we stop pretending that we can’t see Earth’s surface up close with all the tools of the trade, then there are many unique features of plate tectonics worth listing. Just the existence of continents owes itself to plate tectonics, since all that buoyant granite floating around on top of the mantle is primarily created by subduction zone melting. For that matter, subduction-zone volcanism is unique to plate tectonics and Earth, and perhaps if an astronomer were to look for anything on other planets in other solar systems it would be the signs of such volcanism in the atmosphere, e.g., sulfur aerosols [Kaltenegger et al., 2010].

Figure 1: Configuration of present day plate tectonics with indicated directions of motion, on an equal area projection map. Color code for each plate are as follows: Pacific – yellow; Austro-Indian – gold; N. American – light yellow; S. American – red; African – azure; Eurasian – purple; Antarctic – chartreuse; Nazca – indigo; Cocos – violet; Caribbean – light green; Philippine – dark green; Arabian – dark red.

The networks of plate tectonic boundaries, what define plate tectonics in a way, are also unique features (and here’s a gratuitous and probably unnecessary figure of the plates: Fig 1). Subduction zones are where we think the plates are heavy enough to sink into and eventually cool off the mantle, and in the meantime pull the plates behind them. Most of us geodynamicists consider plate motion being largely driven by such subduction or “slab pull” (see the prior blog also http://wp.me/pNy1x-wV). This generally also fits the picture of mantle convection driven by cooling from the surface, as opposed to largely heated from its base. In such surface cooled convection, downwelling of cold material is the dominant circulation. While the mantle is also heated by the core, this is thought to account for at most 20% of heat passing through the mantle [Jaupart et al, 2015]. The slab and cold-downwelling dominated picture of mantle flow also fits the observation of Forsyth and Uyeda [1975] that plates speed are very well correlated with how much slab a plate has; fast plates have a lot of subducting slab, and slow plates have nearly none. There are not many smoking guns in our business, but that one, to me, stands the test of time (well 40 years anyway), and even deserves a figure (Fig 2).

Figure 2 : Percent of plate boundary that is trench or subduction zone versus plate velocity for each plate. (Acronyms are as follows: EUR is the Eurasian plate, NAM is the North American plate, SAM is the South American plate, ANT is the Antarctic plate, AF is the African plate, CAR is the Carribean plate, ARAB is the Arabian plate, IND is the Indo-Australian plate, PHIL is the Philippine plate, NAZ is the Nazca Plate, PAC is the Pacific plate, and COC is the Cocos plate.) The effective trench length corrects for subduction zones whose associated slab-pull forces are opposite in direction and presumably cancel. After Forsyth and Uyeda [1975 ].
Slabs on Earth also drag down surface materials like hydrated and carbonated minerals and sediments (and are the primary driver for volatile exchange between the Earth’s surface and interior). Subduction is also enigmatically one sided; only one plate sinks, while the over-riding plate does not, even if it doesn’t have buoyant continent holding it up.

Subduction zones might also occur on Venus [Schubert and Sandwell, 1995], although this is still debated, but perhaps were generated by plumes similar to what Gerya has proposed to have happened on early Earth [Gerya et al., 2015, and also see the prior blog]. Arcuate structures on Venus that look like trench rises (the part of the subducting plate that bends up before being dragged down) also look two-sided (as pointed out to me by Robert Petersen when I was visiting Scripps a few years back).

On the opposite end of the plates are the mid-ocean ridges, where plate tectonics was effectively first discovered. Although many text books sadly draw mid-ocean ridges as the site where two wheel-like convection cells upwell from the deep mantle, thereby prying apart the plates, that is probably very wrong. Ridges are for the most part passive in that there is no deep buoyant current rising up and pushing them open. Ridges are mostly isostatically floating (most easily inferred because they are nearly invisible in gravity anomaly maps, since the gravitational attraction of surface topographic masses is effectively cancelled by the near-surface mass deficit of their isostatic roots) and so they do not have some very deep buoyant roots pushing them up. Some exceptions to this exist, perhaps most notably in the East African rift, since there might be one of two large deep upwellings there, the other being in the Southwest Pacific. These upwellings have recently been inferred to perhaps help drive plate tectonics by pushing open ridges, although the correlation of plate speed with subduction (referred to above) is hard to get around.

However, ridges are perhaps some of the most unique features of plate-tectonicness. That ridges are generally not the sites of active upwelling but are still focussed and narrow is rather remarkable. From a fluid dynamicist’s perspective, a convecting fluid that is driven by cooling from above (as described above) has only broad passive upwelling, rising in response to cold downwellings dropping into the fluid. Again, this goes with the picture that there is no deep focussed upwelling (mostly) driving apart ridges, but ridges are still narrow. One could argue it’s because melting occurs to lubricate and focus the ridges, which is true; but it’s a chicken or egg question (like the chickens and eggs my fellow blogger is so fond of ) in that melting only occurs because ridges are narrow and focussed, causing shallow mantle to rise up at the spot and undergo localized pressure-release melting.  Something had to get it started.

And then there are strike-slip or transform fault boundaries. Although strike-slip faults like the San Andreas system and Anatolian Fault are some of the more famous of such boundaries, there is a lot of such strike-slip motion in the subduction systems in the Southwest Pacific where the Pacific and Indo-Australian plate are moving with nearly perpendicular trajectories (Fig 3). Strike-slip motion is one of the most unique features of plate tectonics, mainly because it is not something that easily arises from convective motion in the mantle. Slabs are well identified with cold downwellings; and while ridges are mostly passive they still involve shallow upwelling and heat transport out of the mantle. Both slabs and ridges are thus reasonably well associated with mantle convection (though they still have their enigmas, like one-sidedness of subduction, and narrowness of passive upwelling). Any terrestrial planet that is undergoing mantle convection will have something that looks like upwelling and downwellings of some sort. But strike-slip motion doesn’t appear to do anything for convection or cooling the planet since it doesn’t transport mass or heat vertically. However, a big fraction (maybe up to 50%, though this depends on the plate-tectonic frame of reference) of plate tectonic energy is in this strike-slip motion. So strike slip motion is not only one of those unique “scars” by which you’d recognize plate tectonics, it’s also hard to understand why it exists at all.

In summary I would say (just to be simple minded about it), that the main features of plate tectonicness, are (1) one-sided subduction of tectonic plates that involves entraining surface material (and from this the other unique feature of continental crust pops out), (2) focussed but passive divergent zones or spreading centers, and (3) significant strike-slip motion. If we want to recognize plate tectonics in all its glory on another planet or identify how, when and why it started on Earth, we need to find these features (somehow). But then it also helps to understand why we have these features at all.

What are the ingredients to get a plate-like planet?

One of the grand challenges of our business is to answer the question of why does Earth have plate tectonics, or more specifically why does mantle convection on Earth look like plate tectonics at the surface? This question has fueled work and debate since the 1970s for almost as long as the theory of plate tectonics has been around [e.g., see seminal works of Turcotte and Oxburgh, 1972, Kaula, 1980, Hager and O’Connell, 1981].  The physics of what causes the plates to acquire their unique features has been a huge challenge in our field, with many players, and something that has consumed the author’s professional career longer than he cares to admit. Much of the physics is probably too boring to survey in detail, and interested readers can always dig into it with recent reviews [to be self serving, I’ll point the reader at Bercovici et al., 2015]. But I will at least try to give the good parts version of where we stand now (understandably from my biased perspective).

Getting softer

A common aspect of all plate tectonic boundaries is that they are focused and narrow relative to the size of the plates themselves, which move as mostly big solid blocks or (hence the name) plates. Most of the “action” is of course at the plate boundaries, i.e., almost all the deformation in the lithosphere is localized at these plate boundaries. One might think (and it’s sometimes been claimed) that such narrow deformation is just a big crack; however cracking and failure only goes so deep, about ten kilometeres, while plates and the lithosphere are (most of the time) about hundred kilometers thick; thus most of the deeper lithosphere has to sustain focused and intense deformation without just simple cracking.

Localizing deformation is then usually attributed to some sort of self-weakening feedback effect, i.e., if we deform something by bending or twisting it, it gets weaker where it’s being most deformed; the weak spot focuses the deformation since it’s easier to twist, bend, etc, which makes it deform faster, which makes it softer and the deformation more localized, and so on. Some materials naturally act this way; they get softer once they’re deformed, e.g., many food products like mustard (which is strong enough when resting to retain bumps on the surface, but once it’s stirred or squeezed out of a bottle it moves easily). This is referred to as plastic behavior, and is described by a relatively simple physical law, i.e., there is no deformation if the stress acting on the medium is below a yield or critical value, but it flows like a fluid if stress is above the yield value. This law is easily employed in models of lithospheric deformation and has the attraction of allowing weakening in the deep lithosphere, but still looks a bit like failure or “cracking” conditions in the shallow lithosphere. Many models of convection using this have been very successful at obtaining plate-like behavior, where narrow regions of the cold top layer of the convecting system soften and deform, while large regions of lithosphere that are below the yield stress remain block-like (Fig 4). Self-softening is even more profound if the strength (i.e, resistant stress) of the deforming zones – not just the viscosity – actually drops or vanishes with faster deformation, much like stick-slip behavior describing failure and sliding [an idea I’ve pushed off and on for 20 years, e.g., Bercovici, 1993, 1995a], although this model is harder to use in fluid mechanical models of convection [Tackley, 1998, 2000].

Figure 4A: Three-dimensional spherical shell convection with a plastic-type lithospheric rheology from two different studies of van Heck and Tackley [2008] (top) and Foley and Becker [2009] (bottom) showing plate-like behavior. Left panel shows surface viscosity with velocity vectors super-imposed. Right panels show isothermal surfaces and in particular cold downwellings. Note the passive divergent and rheological weak zone forming mid-way between the two major downwelling regions.
Fig. 4B: Surface viscosity with velocity vectors super-imposed.
Fig 4C: Isothermal surfaces.

Such self-softening behavior is also necessary to explain the occurrence of the enigmatic strike-slip behavior plates. Generally this takes a mathematical explanation, but perhaps the best way to explain it is that the self-softening promotes lubricated strike-slip tracks to guide motion of lithosphere from a ridge to a subduction zone [Bercovici, 1995b]. Without these slippery tracks, the plate would have to deform enormously while trying to slide past neighboring lithosphere, and would probably then just seize up.

Lasting scars

Although such self-softening rheologies have gotten us quite far, they have some failings. First, the yield stresses needed to get plate-like behavior have typically been unusually low (much lower than experiments on rock strength would imply). Moreover, the weak zones or plate boundaries would be extremely ephemeral; the plastic softening occurs only so long as that region is being deformed, but once that stops it goes back to being stiff again. In essence the model plate boundaries only exist while deforming, but instantaneously vanish when they stop. As pointed out by Mike Gurnis and colleagues (about 20 years ago, probably even in response to some of my papers!), this isn’t very plate-like [Zhong et al., 1998, Gurnis et al., 2000].

Plate margins and faults tend to last a long time after going dormant, and often provide an inherited weak spot or scar, like a suture zone, that can be re-used. One of the toughest nuts to crack in our business is understanding how subduction zones start at all. Getting a full-blow subduction zone to develop in thick, cold and incredibly strong lithosphere is essentially impossible and one has to torture pristine lithosphere (with imposed rifting, or loading with sediments and/or thick passive margins) to get it to cough up a subduction zone. However it’s perhaps more likely that subduction (in its full surface-entraining glory) initiates off dormant plate boundaries and faults, which act as a weak zone on which to nucleate sinking of a plate edge [Toth and Gurnis, 1998, Gurnis et al., 2000, Lebrun et al., 2003, Hall et al., 2003]. Thus the persistence of weak zones and plate boundaries after they go to sleep is a crucial feature of plate tectonics.

Getting a self-weakening mechanism to operate in the strongest, deeper lithosphere in a way that leaves a semi-permanent scar is thus an important goal (in my view). And in that regard the Earth has left us an important clue. But to understand it we have to abandon our global mantle convection view of the Earth and dive down into the microscopic or grain scale perspective. (Ok, yes, those who bore witness to my 2014 Birch lecture probably recall the cheesy movie I made to segue into this topic.) 

The world in a grain of….

Well, ok, William Blake’s quote is actually about a grain of sand:

To see a world in a grain of sand
And heaven in a Wild Flower
Hold infinity in the palm of your hand
And eternity in an hour.

Maybe a grain of olivine isn’t as poetic, although anyone who has been to Papakōlea Beach on the Big Island of Hawaii might disagree. But getting past my digression, Blake’s quote is in a way profound, since the scales of time and distance of plate tectonics and planetary evolution might depend on the physics at the scale of a grain. But then again, Blake is the go-to poet for scientific quotes, so don’t be impressed by my literary allusions (and no Yale jokes, please).

The strongest part of the lithosphere that has to be softened, which is thus the plate tectonic bottle-neck, is in my view the cold ductile region, below where failure and frictional sliding become ineffective. This is at about 50-100km in depth (more or less) and ranging mostly around temperatures of 1000K. It turns out that rocks in deformation zones exhumed from these depths and temperatures exhibit very localized deformation and at the same time unusual grain-size reduction; these are referred to as mylonites (Fig 5). The origin and formation of mylonites has been a subject of research and debate for many years, and that work is, in my opinion, of enormous importance to understanding the origin and cause of plate tectonics. There are many heroes of this field, from people doing laboratory rock deformation experiments to field studies of deformation belts, and the literature is extensive. However I would be remiss in not mentioning a few, although this will emphasize those who have been most influential in my own thinking. These include (of course) my Yale colleague Shun Karato, as well as Greg Hirth, Dave Kohlstedt, Brian Evans, Phil Skemer, Lars Hansen, and the work of Marco Herwegh and his (former) student Jolien Linckens [e.g., Karato et al., 1980, Karato, 1989, Jin et al., 1998, Hirth and Kohlstedt, 2003, Warren and Hirth, 2006, Evans et al., 2001, Austin and Evans, 2007, Skemer et al., 2010, Hansen et al., 2012, Herwegh et al., 2011, Linckens et al., 2015]. I’ll also refer the interested reader to my 2015 review paper [Bercovici et al., 2015] or my most recent (slightly less readable) paper [Bercovici and Ricard, 2016] for many references on these studies. But I will say, overall, that the experimental and field-based rock deformation researchers cannot possibly get enough credit for doing the hard work in revealing critical clues about how the Earth works.

Figure 5: A wonderful example of a peridotitic or mantle mylonite, from Lars Hansen’s website and from work done for Hansen et al. [2013]. The bands of small grains wrapping around the large olivine porphyroblasts are the mylonites having undergone extreme deformation and grain-reduction.
The bottom line is that, in my take, the origin of mylonites is the origin of plate boundaries, for the most part. Having said that though, not everyone even agrees on the origin of mylonites. In principle a self-softening feedback would occur in mylonites because deformation causes grains to shrink, and the shrinking grains would make the material softer, which makes the deformation faster and more focused, which makes the grains shrink faster yet, and so on. But that’s easier said than done.

While grains are big (relative to a size depending on stress and temperature) they deform by dislocation creep whose rheology could care less about grain size, although the accumulation of dislocations in the minerals promotes smaller grains to form and dominate the mixture. Once the grains get small enough they deform by diffusion creep, whose rheology cares deeply (just to anthropomorphize rheologies a bit more) about grain size in that smaller grains lead to weaker material (for rocks at some depth, but not for everything; it’s opposite in cold metals); see Fig 6. This would give the requisite self-softening, but the cause for grain-reduction while the grains are small and no longer in dislocation creep (where dislocations promoted grain reduction) is unclear. Without this means of grain reduction, the grains would swell by normal grain-growth or coarsening (much the way bubbles in foam get bigger as the system’s surface energy goes to a minimum), causing the material to get stronger.


There is, however, ample evidence that the presence of polycrystalline mixtures, like the mixture of olivine and pyroxene (and other components) in mantle-lithosphere peridotite, promotes the formation of mylonites by a sort of mutual Zener pinning effect (where grain growth in one component like olivine is blocked by the presence of the other component like pyroxene, and vice versa); this effect can help fix and even possibly reduce grain-sizes in the diffusion creep field [e.g., Warren and Hirth, 2006, Herwegh et al., 2011, Linckens et al., 2015, Bercovici and Ricard, 2012]. Likewise, there may be other effects allowing grains to stay small and shrink in the diffusion creep regime, like nucleation of new small grains during cooling, reactions or phase changes [e.g., Linckens et al., 2011]. And probably the most effective mechanism is some combination of all these effects.

The attractiveness of this mylonite or grain-reduction idea (or as I and my colleague Yanick Ricard have called it “grain-damage” [Bercovici and Ricard, 2012, 2013, 2016], but no one ever seems to acknowledge the bad joke, perhaps for good reason), is that in a way we can eat our cake and have it, too. (I know the actual saying is reversed, but it never made sense to me; of course I can have a cake and eat it, too; but I can’t eat a cake and then have it, right? This is important.) That is, the grain-reduction allows a positive self-weakening feedback (admittedly with all the difficulties of explaining mylonites), which causes narrow localized deformation in the lithosphere yielding plate boundaries and separating strong plates. But, if deformation stops then weak zones do not automatically vanish but persist as long as the mylonites or small grains last. The grains will grow back eventually and, in effect, heal the weak zone; but if the rock is a mixture of different minerals, which it obviously is, then the mutual Zener pinning effect blocks grain-growth and drastically slows down this healing rate, potentially for geologically very long times [Bercovici and Ricard, 2012, 2014].

There is still a lot of this picture to be worked out and things we don’t understand, but the general picture gives us a framework with which to start, and most importantly is grounded (or we try to be as grounded as we can) in the important observational work on rock physics and mylonites, from the lab and the field.

What about water?

The co-existence of plate tectonics and water is probably no coincidence, although we also have only one data point on which to base this statement: the Earth. Nevertheless, water is often heralded as a necessary condition for plate tectonics. It might very well be plate tectonics and water are necessary conditions for each other, which always brings up the chicken-or-egg question of which came first? But we’ll save that question for a bit later.

Water is probably necessary to help keep subduction zones going and is often invoked as a way to lubricate subduction with sediments. Moreover, as my fellow blogger Taras Gerya has argued [Gerya et al., 2008], water helps make subduction one-sided; in particular, hydrous melting of mantle above a subducting slab helps keep that mantle corner or wedge too weak to drag down the over-riding plate. However, water’s role in actually weakening the lithosphere and promoting a self-softening feedback leading to plate boundaries is not so obvious. When the lithosphere is formed at mid-ocean ridges, partial melting extracts water and leaves the lithosphere fairly dried out [Hirth and Kohlstedt, 1996] hence Earth’s lithospheric mantle is perhaps no wetter than on Venus. Resupplying water from the surface into the deeper lithosphere is difficult since it involves pushing water against very high lithostatic pressures, although there may be ways to do it with deep cracking under thermal stresses [Korenaga, 2007].

As far as the grain-damage or mylonite mechanism (discussed above) goes, there is a different water story, which does not depend on deep ingestion of water. The grain damage mechanism works mostly as a competition between damage under deformation (i.e., creating smaller grains) against healing under normal grain-growth. Grain-growth is slower at cooler temperatures, and faster at higher temperatures. The efficiency of damage is also stronger at cooler temperatures, and less at higher ones. Thus the damage-to-healing ratio increases with low temperature and drops at higher temperature. Water’s role in this is that, in its liquid state, it promotes the geological carbon cycle by which atmospheric CO2 is drawn down, keeping the surface at a cooler temperature, which then facilitates damage and suppresses healing. Water does not (in this picture) cause the self-softening feedbacks and formation of plate boundaries, but simply helps keep the temperature cool enough for this feedback to exist. [Landuyt and Bercovici, 2009, Foley et al., 2012],

This idea of a climate-controlled plate-generating mechanism also helps explain why our ostensible twin Venus does not have plate tectonics; its surface is so hot that it promotes healing over damage and thus plate boundaries don’t last if they form at all. But the climate control of plate tectonics is not unique to the grain-damage model and works well even in models of a lithosphere with a plastic rheology; in essence the hotter surface leads to a lighter or less negatively buoyant lithosphere, which then doesn’t have enough weight to exceed the yield stress [e.g. Lenardic et al., 2008].

If water and climate exert some control on generating and maintaining plate tectonics, and plate tectonics is necessary to drive the long-term carbon cycle that keeps CO2 levels low and hence a clement climate, then water and plate-tectonics are mutually supporting [e.g., Sleep and Zahnle, 2001, Sleep, 2015, Foley, 2015, Foley and Driscoll, 2016]. And this gets us back to the question of what came first? Water or plate tectonics? I’ll speculate more on this in the next section.

What can we say about how plate tectonics began?

One of the more important questions is, what did mantle convection look like after the magma ocean? However, I’m going to leave this question of the magma ocean and conditions immediately after it to the next blogger whose cool middle name actually means “Magma Ocean” (she gets the last word, so I’m sure retribution is coming my way).

In the early Archean world, it’s likely that the Earth’s surface was much hotter, with a thick CO2 rich atmosphere not so different than Venus’s atmosphere now. The mantle was also younger and hotter with more radiogenic heat sources blasting away at mantle rocks. Thus if the Earth and mantle were hotter and had more energy to lose, wouldn’t plate tectonics be faster? However, the opposite is possibly the case, for several reasons. The higher temperatures and subsequent greater degree of melting could have caused the lithosphere to be, drier, stronger (relative to the mantle) and, with the extra production of buoyant crust, harder to deform and subduct [Davies, 1992, 2007, Korenaga, 2006, 2013, van Hunen and Moyen, 2012]. This might suggest it would be very hard to get plate tectonics until late in the Archean or even Proterozoic, as some researchers have proposed (see our opening blog for discussion).

However, there are some lines of evidence, including the discovery of zircons from Australia, suggesting that at least something like subduction – which is the main generator for granites and hence zircons – existed more than 4 billion years ago [Harrison et al., 2005, Shirey et al., 2008, Hopkins et al., 2010]. The evidence for this is sparse and controversial and only in one location. Likewise, there is ample evidence that the Earth underwent a transition to more global plate tectonics by 3-2.7 billion years ago [e.g., Condie and Kröner, 2008, Shirey and Richardson, 2011]. However, I will leave it to our blogger-in-chief’s contribution to cover the geological evidence for when plates started. My real goal is to see what we can say about how plate tectonics began, but using the physics learned from trying to explain why we have plate tectonics at all.

If plate tectonics started 4Ga and the record afterwards just happens to be very spotty [Korenaga, 2013] then that is one mystery to solve. It is possible that despite the hot and dry lithosphere, the mantle was also dried out, perhaps after expelling water due to early degassing; thus the mantle was not so much weaker than the lithosphere and mantle convection could drag down the lithosphere and get subduction going. After subduction started, water was possibly re-ingested into the mantle, causing it to be much weaker than the lithosphere, thereby slowing down plate tectonics enough to keep the mantle from cooling off too fast [see Korenaga, 2013].

It is also possible that plate tectonics didn’t start all at once, as proposed by Condie and Kröner [2008], but started off locally and only became global by about 3-2.7 billion years (we can just call it 3Gyrs). The grain-damage and mylonite physics I discussed above possibly helps explain this scenario [Bercovici and Ricard, 2014]. In particular, we proposed that a convective instability like a mantle downwelling, or drip or protosubduction initiated and induced a damage zone or scar by virtue of grain reduction. Such drips and isolated downwellings tend to be chaotic, intermittent or ephemeral, and might vanish and a new downwelling appear elsewhere. However the damage zones left by the first downwellings would not heal quickly, and then would reactivate and be the location of deformation once new downwellings formed nearby; motion on the old damage zones might involve spreading or strike-slip motion, depending on their orientation. But the main point is that revived motion keeps the old damage zones weak and localized. As downwellings form and vanish they leave behind new scars or weak zones each inherited to accommodate lithospheric motion. Eventually these weak zones accumulate enough to make a network of complete plate boundaries, wherein passive spreading and strike-slip margins are all driven just by subduction. This, the story goes, could occur in about 1 Gyrs from 4Ga to 3Ga, largely limited by the formation and demise of downwellings. In the end, plate tectonics arises by the formation, accumulation and inheritance of weak grain-damaged zones (Fig. 7  and related movies).

Figure 7: A calculation of pressure-driven flow in a thin 2-D horizontal-layer model of the lithosphere, where the pressure low is akin to a single subduction zone (see schematic frame). The rheology of the layer is governed by the grain evolution, damage and pinning model of Bercovici and Ricard [2012 , 2013 ]. In this case the low pressure zone P is imposed and then rotated about a vertical axis by 90 degrees three times (with roughly 10Myrs dimensional time between rotations to develop damaged weak zones) as an idealization of intermittent and chaotic subduction during the early Archean. The bands of damage induced by the pressure low from a previous orientation are long-lived, inherited and amplified by the lithospheric flow of the next orientation, resulting in localized but passive bands of strike-slip vorticity  and positive divergence S (red and yellow contours; convergence with S < 0 , indicated by blue contours, is actively driven by the pressure low). Thus a complete plate arises with a contiguous weak plate boundary (indicated by viscosity  ) while only being driven by subduction, which is the final state shown (with final dimensionless time indicated on the P frame). Dimensionless extrema for contours or vectors are indicated below each frame, save P which is always between 0 and 1. This model is proposed to explain the emergence of plate tectonics in the Archean, from proto-subduction 4Ga to global tectonics by ~3Ga. Adapted from Bercovici and Ricard [2014], which also gives further details of the calculation.
Movie 1 still
A representation of the model of Bercovici & Ricard [2014], discussed in Fig 7, but wrapped on rotating sphere. A low pressure band, as a proxy for the suction of a subducting slab, initiates convergence (negative divergence or the blue band on the “divergence” sphere) but then vanishes and a new pressure zone occurs perpendicular to the last one, and the sequence is repeated twice more. For the Earth-like case with a cool surface, damaged, weak zones accumulate, as shown in the “viscosity” sphere, and strike-slip bands of vorticity develop; in the end there is a convergence zone driven by the final low-pressure proxy-slab, with a passive divergence zone and strike-slip margins, as shown in Figure 7. The movies for the divergence, viscosity, and vorticity fields are shown below.

Movie 2 still
For the Venus case, the temperature is higher, damage is weaker and healing stronger, and so the low-viscosity weak zones do not form or last or accumulate as well as in the Earth case, leaving the final flow field dominated by convergence, with little if any localized divergence or strike-slip motion.  The movies for the divergence, viscosity, and vorticity fields are shown below.

In contrast, Venus always being too hot would not support weak damaged zones long enough to be inherited, since healing would be faster and damage weaker. Hence while Venus might have something like subduction, it would not have accumulated the other passive plate boundaries.

However it’s also possible that the first motion was not a spontaneous drip or proto-downwelling. My fellow bloggers, Taras Gerya and Bob Stern and colleagues have proposed that strong mantle plumes caused doming and initiated subduction off the edge of these domes [Gerya et al., 2015, and see prior blog]. This plume-initiated subduction model is appealing in that this is possibly what Venus is doing now, i.e., at the possible occurrence of subduction at the edge of large coronae and other large dome-like features. Of course this is discussed in the previous blog so there is no point in my repeating it here. However, for plate tectonics to achieve a global network on Earth, weak zones would need to accumulate, especially to permit passive spreading and strike-slip boundaries, and thus something like the cycle of damage and inheritance would need to occur. Nevertheless, these recent models are complementary and very much build on each other.

The independent evolutionary paths of the terrestrial planets, especially Venus and Earth, might also occur because the physics of how they evolve is path-dependent. In other words, there may be a hysteresis effect where there are entirely different evolutionary trajectories depending on how a planet starts its journey through time [Sleep, 2000, 2015]. For example if a planet starts slightly too hot, it might never reach a plate-tectonic pathway or branch in its development. Various approaches to understanding the physics dictating the generation of plate tectonics has supported this inference, including plastic-lithosphere approaches [see Weller and Lenardic, 2012, O’Neill et al., 2016], and the grain damage/mylonite theory [Bercovici and Ricard, 2016].

I’ll finally loop back on this question of what came first, water or plate tectonics. This is also related to the divergent evolutionary paths of Earth and Venus [see recent review by Foley and Driscoll, 2016]. The evidence from zircons in Australia not only points to granites and subduction, but also the presence of water more than 4Gyrs ago [Mojzsis et al., 2001, Valley et al., 2002], and so subduction and water were possibly both around to some extent, although perhaps not globally. If we believe that a cool enough climate was needed for plates to form (or weak zones to accumulate) and that plate tectonic recycling was necessary to draw down CO2 to allow liquid water to form, then they would make a mutually-promoting feedback. But it’s unlikely the feedback could start without some seed or kick in the right direction. My own best guess is that liquid water probably came first, if even in small quantities. If there was no liquid water and it was all vapor (as likely on Venus early on) then it is difficult to efficiently draw down CO2 without some hydrological cycle, even if subduction and rock resurfacing somehow got going on its own. But even with a tiny amount of water and a weak hydrological cycle, CO2 could be drawn down, perhaps even by reacting with basalt from mantle melting and heat-pipe heat transport and volcanism [Moore and Webb, 2013], or the last crustal remains of the magma ocean [see Elkins-Tanton, 2012]. This might have caused enough cooling to promote more liquid water and thus more CO2 draw-down, while then promoting better conditions for plate tectonics to get a foot-hold, which in turn would enhance the rock cycling, CO2 draw down and further growth of the oceans. Whether this took as long as the slow accumulation of plate tectonics is a big question, and then again both the ocean and full plate tectonics might have come into existence quickly and early. Even so, the critical issue for Venus is that it probably never had a kernel of liquid water to get this feedback and machinery going.

We have learned a lot about the cause for plate tectonics by looking at the current Earth and considering the physics of what causes plate tectonics to exist, and what are the conditions by which it would not exist as on Venus. We can apply this physics within a limited way to say how plate tectonics started. But until we know much, much more about the early Earth, or perhaps Venus as an analog for the early Earth, we can only speculate. Currently it is extremely difficult to explore Venus and especially its surface geology. But then again most of the rock record of the early Archean Earth is lost, the culprit for which is the very thing we wish to understand, plate tectonics. In the end, to understand how and why plate tectonics started, it would probably be easier to land on Venus than to go back in time.


N. Austin and B. Evans. Paleowattmeters: A scaling relation for dynamically recrystallized grain size. Geology, 35:343–346, 2007.

D. Bercovici. A simple model of plate generation from mantle flow. Geophys. J. Int., 114: 635–650, 1993.

D. Bercovici. A source-sink model of the generation of plate tectonics from non-newtonian mantle flow. J. Geophys. Res., 100:2013–2030, 1995a.

D. Bercovici. On the purpose of toroidal flow in a convecting mantle. Geophys. Res. Lett., 22:3107–3110, 1995b.

D. Bercovici and Y. Ricard. Mechanisms for the generation of plate tectonics by two phase grain-damage and pinning. Phys. Earth Planet. Int., 202-203:27–55, 2012.

D. Bercovici and Y. Ricard. Generation of plate tectonics with two-phase grain-damage
and pinning: Source–sink model and toroidal flow. Earth and Planetary Science Letters,
365:275 – 288, 2013.
D. Bercovici and Y. Ricard. Plate tectonics, damage and inheritance. Nature, 508:513–
516, 2014.

D. Bercovici and Y. Ricard. Grain-damage hysteresis and plate-tectonic states. Phys. Earth Planet. Int., 253:31–47, 2016.

D. Bercovici, P. J. Tackley, and Y. Ricard. The Generation of Plate Tectonics from Mantle Dynamics. In David Bercovici (Gerald Schubert editor-in-chief), editor, Treatise on Geophysics, volume 7, chapter 7, pages 271 – 318. Elsevier, 2 edition, 2015. doi:

K. Condie and A. Kröner. When did plate tectonics begin? Evidence from the geologic
record. In K. Condie and V. Pease, editors, When Did Plate Tectonics Begin on Planet
Earth?, pages 281–294. Geological Society of America, 2008. Special Paper 440.

J. E. P. Connerney, M. H. Acu˜na, P. J. Wasilewski, N. F. Ness, H. R`eme, C. Mazelle,
D. Vignes, R. P. Lin, D. L. Mitchell, and P. A. Cloutier. Magnetic lineations in the
ancient crust of Mars. Science, 284(5415): 794–798, 1999. doi: 10.1126/science.284.

J. E. P. Connerney, M. H. Acu˜na, N. F. Ness, G. Kletetschka, D. L. Mitchell, R. P. Lin,
and H. Reme. Tectonic implications of mars crustal magnetism. Proceedings of the
National Academy of Sciences of the United States of America, 102(42):14970–14975,
2005. doi: 10.1073/pnas.0507469102.

G. Davies. On the emergence of plate tectonics. Geology, 20:963–966, 1992.

G. Davies. Thermal evolution of the earth. In D. Stevenson (G. Schubert, chief editor), editor, Treatise on Geophysics, volume 9, Evolution of the Earth, pages 197–216. Elsevier, Amsterdam, 2007.

C. Dumoulin, D. Bercovici, and P. Wessel. A continuous plate-tectonic model using geophysical data to estimate plate margin widths, with a seismicity based example. Geophys. J. Int., 133:379–389, 1998.

L. T. Elkins-Tanton. Magma oceans in the inner solar system. Annual Reviews of Earth and Planetary Sciences, 40(1):113–139, 2012. doi: 10.1146/annurev-earth-042711-105503.

B. Evans, J. Renner, and G. Hirth. A few remarks on the kinetics of static grain growth in
rocks. Int. J. Earth Sciences (Geol. Rundsch.), 90:88–103, 2001.

B. Foley and T. Becker. Generation of plate-like behavior and mantle heterogeneity
from a spherical, visco-plastic convection model. Geochem., Geophys., Geosys., 10
(8):Q08001, doi:10.1029/2009GC002378, 2009.

B. J. Foley. The role of plate tectonic-climate coupling and exposed land area in the
development of habitable climates on rocky planets. The Astrophysical Journal, 812(1):
36, 2015.

B. J. Foley and P. E. Driscoll. Whole planet coupling between climate, mantle, and core:
Implications for rocky planet evolution. Geochemistry, Geophysics, Geosystems, 2016. ISSN 1525-2027. doi: 10.1002/2015GC006210.

B. J. Foley, D. Bercovici, and W. Landuyt. The conditions for plate tectonics on superearths: Inferences from convection models with damage. Earth and Planetary Science Letters, 331-332:281–290, 2012.

D. Forsyth and S. Uyeda. On the relative importance of the driving forces of plate motion. Geophys. J. R. Astr. Soc., 43:163–200, 1975.

T. V. Gerya, J. A. Connolly, and D. A. Yuen. Why is terrestrial subduction one-sided?
Geology, 36(1):43–46, 2008. doi: 10.1130/G24060A.1.

T. V. Gerya, R. J. Stern, M. Baes, S. V. Sobolev, and S. A. Whattam. Plate tectonics on the earth triggered by plume-induced subduction initiation. Nature, 527(7577):221–225, 11,2015

M. Gurnis, S. Zhong, and J. Toth. On the competing roles of fault reactivation and brittle
failure in generating plate tectonics from mantle convection. In M. A. Richards, R. Gordon, and R. van der Hilst, editors, History and Dynamics of Global Plate Motions,
Geophys. Monogr. Ser., volume 121, pages 73–94. Am. Geophys. Union, Washington,
DC, 2000.

B. Hager and R. O’Connell. A simple global model of plate dynamics and mantle convection. J. Geophys. Res., 86:4843–4867, 1981.

C. E. Hall, M. Gurnis, M. Sdrolias, L. L. Lavier, and R. D. Mueller. Catastrophic initiation of subduction following forced convergence across fracture zones. Earth Planet. Sci. Lett, 212:15–30, 2003.

L. Hansen, M. Zimmerman, A. Dillman, and D. Kohlstedt. Strain localization in
olivine aggregates at high temperature: A laboratory comparison of constant-strain rate
and constant-stress boundary conditions. Earth and Planetary Science Letters,
333–334(0):134 – 145, 2012. doi: 10.1016/j.epsl.2012.

L. N. Hansen, M. J. Cheadle, B. E. John, S. M. Swapp, H. J. B. Dick, B. E. Tucholke, and
M. A. Tivey. Mylonitic deformation at the kane oceanic core complex: Implications
for the rheological behavior of oceanic detachment faults. Geochemistry, Geophysics,
Geosystems, 14(8):3085–3108, 2013. doi: 10.1002/ggge.20184.

T. M. Harrison, J. Blichert-Toft, W., Müller, F. Albarede, P. Holden, and S. J. Mojzsis. Heterogeneous Hadean Hafnium: Evidence of Continental Crust at 4.4 to 4.5 Ga. Science, 310(5756):1947–1950, 2005. doi: 10.1126/science.1117926.

M. Herwegh, J. Linckens, A. Ebert, A. Berger, and S. Brodhag. The role of second phases for controlling microstructural evolution in polymineralic rocks: A review. Journal of Structural Geology, 33(12):1728 – 1750, 2011. doi: 10.1016/j.jsg.

G. Hirth and D. Kohlstedt. Water in the oceanic upper mantle: implications for rheology,
melt extraction and the evolution of the lithosphere. Earth Planet. Sci. Lett., 144:93–
108, 1996.

G. Hirth and D. Kohlstedt. Rheology of the upper mantle and the mantle wedge: a view
from the experimentalists. In J. Eiler, editor, Subduction Factor Mongraph, volume 138,
83–105. Am. Geophys. Union, Washington, DC, 2003.

M. D. Hopkins, T. M. Harrison, and C. E. Manning. Constraints on hadean geodynamics
from mineral inclusions in >4Ga zircons. Earth and Planetary Science Letters, 298
(3–4): 367 – 376, 2010. doi: 10.1016/j.epsl.2010.08.010.

C. Jaupart, S. Labrosse, F. Lucazeau, and J.-C. Mareschal. Temperatures, heat and energy in the mantle of the earth. In D. Bercovici (G. Schubert, chief editor), editor, Treatise on Geophysics, volume 7, Mantle Dynamics, pages 223–270. Elsevier, New York, 2nd edition, 2015.

D. Jin, S. Karato, and M. Obata. Mechanisms of shear localization in the continental lithosphere: Inference from the deformation microstructures of peridotites from the Ivrea
zone, northwestern Italy. J. Struct. Geol., 20:195–209, 1998.

L. Kaltenegger, W. G. Henning, and D. D. Sasselov. Detecting volcanism on extrasolar
planets. The Astronomical Journal, 140(5):1370, 2010.

S. Karato. Grain growth kinetics in olivine aggregates. Tectonophysics, 168:255–273,

S. Karato, M. Toriumi, and T. Fujii. Dynamic recrystallization of olivine single crystals
during high temperature creep. Geophys. Res. Lett., 7:649–652, 1980.

W. Kaula. Material properties for mantle convection consistent with observed surface
fields. J. Geophys. Res., 85:7031–7044, 1980.

J. Korenaga. Archean geodynamics and the thermal evolution of earth. In K. Benn, J.-C.
Mareschal, and K. Condie, editors, Archean Geodynamics and Environments, volume
164, pages 7–32. AGU Geophysical Monograph Series, 2006.

J. Korenaga. Thermal cracking and the deep hydration of oceanic lithosphere: A key to the generation of plate tectonics? J. Geophys. Res., 112, 2007. doi:10.1029/2006JB004502.

J. Korenaga. Initiation and evolution of plate tectonics on earth: Theories and
observations. Ann. Rev. Earth Planet. Sci., 41:117–151, 2013. doi: 10.1146/

W. Landuyt and D. Bercovici. Variations in planetary convection via the effect of climate
on damage. Earth Planet. Sci. Lett., 277:29–37, 2009.

J.-F. Lebrun, G. Lamarche, and J.-Y. Collot. Subduction initiation at a strike-slip plate
boundary: The Cenozoic Pacific-Australian plate boundary, south of New Zealand. J.
Geophys. Res., 108, 2003. doi: 10.1029/2002JB002041.

A. Lenardic, M. Jellinek, and L.-N. Moresi. A climate change induced transition in the
tectonic style of a terrestrial planet. Earth Planet. Sci. Lett., 271:34–42, 2008.

J. Linckens, M. Herwegh, O. Müntener, and I. Mercolli. Evolution of a polymineralic
mantle shear zone and the role of second phases in the localization of deformation. J.
Geophys. Res., 116:B06210, 2011. doi: 10.1029/2010JB008119.

J. Linckens, M. Herwegh, and O. M¨untener. Small quantity but large effect —how minor phases control strain localization in upper mantle shear zones. Tectonophysics, 643: 26–43, 3 2015. doi: http://dx.doi.org/10.1016/j.tecto.2014.12.008.

S. J. Mojzsis, T. M. Harrison, and R. T. Pidgeon. Oxygen-isotope evidence from ancient
zircons for liquid water at the Earth’s surface 4,300 Myr ago. Nature, 409(6817):178–
181, 2001.

W. B. Moore and A. A. G. Webb. Heat-pipe earth. Nature, 501(7468):501–505, 09 2013.

C. O’Neill, A. Lenardic, M. Weller, L. Moresi, S. Quenette, and S. Zhang. A window
for plate tectonics in terrestrial planet evolution? Physics of the Earth and Planetary
Interiors, 255:80 – 92, 2016. ISSN 0031-9201. doi: http://dx.doi.org/10.1016/j.pepi.
A. Rozel, Y. Ricard, and D. Bercovici. A thermodynamically self-consistent damage equation for grain size evolution during dynamic recrystallization. Geophys. J. Int., 184(2):719–728, 2011. doi: 10.1111/j.1365-246X.2010.04875.x.

G. Schubert and D. Sandwell. A global survey of possible subduction sites on Venus.
Icarus, 117(1):173 – 196, 1995. ISSN 0019-1035. doi: 10.1006/icar.1995.1150.

S. Shirey, B. Kamber, M. Whitehouse, P. Mueller, and A. Basu. A review of the isotopic
and trace element evidence for mantle and crustal processes in the Hadean and Archean:
Implications for the onset of plate tectonic subduction. In K. Condie and V. Pease,
editors, When Did Plate Tectonics Begin on Planet Earth?, pages 1–29. Geological
Society of America, 2008. Special Paper 440.

S. B. Shirey and S. H. Richardson. Start of the Wilson Cycle at 3 Ga shown by diamonds
from subcontinental mantle. Science, 333:434–436, 2011.

P. Skemer, J. M. Warren, P. B. Kelemen, and G. Hirth. Microstructural and rheological
evolution of a mantle shear zone. J. Petrol., 51:43–53, 2010. doi:10.1093/petrology/

N. Sleep. Evolution of the mode of convection within terrestrial planets. J. Geophys. Res. 105:17,563–17,578, 2000.

N. Sleep. Evolution of the Earth: Plate tectonics through time. In D. Stevenson (G. Schubert, chief editor), editor, Treatise on Geophysics, volume 9, chapter 6, pages 145 – 172. Elsevier, Oxford, 2 edition, 2015. doi: http://dx.doi.org/10.

N. Sleep and K. Zahnle. Carbon dioxide cycling and implications for climate on ancient
Earth. J. Geophys. Res., 106:1373–1399, 2001.

P. Tackley. Self-consistent generation of tectonic plates in three-dimensional mantle convection. Earth Planet. Sci. Lett., 157:9–22, 1998.

P. Tackley. Self-consistent generation of tectonic plates in time-dependent, three dimensional mantle convection simulations, 2. Strain weakening and asthenosphere.
Geochem. Geophys. Geosystems (G3), 1:2000GC000043, 2000.

G. Toth and M. Gurnis. Dynamics of subduction initiation at preexisting fault zones. J.
Geophys. Res., 103:18053–18067, 1998.

D. Turcotte and E. Oxburgh. Mantle convection and the new global tectonics. Ann. Rev.
Fluid Mech., 4:33–66, 1972.

D. Turcotte, G. Morein, D. Roberts, and B. Malamud. Catastrophic resurfacing and
episodic subduction on venus. Icarus, 139(1):49 – 54, 1999. doi:

D. L. Turcotte. An episodic hypothesis for Venusian tectonics. Journal of Geophysical
Research: Planets, 98(E9):17061–17068, 1993. doi: 10.1029/93JE01775

J. W. Valley, W. H. Peck, E. M. King, and S. A. Wilde. A cool early Earth. Geology,
30(4):351–354, 2002. doi: 10.1130/0091-7613 (2002)

H. van Heck and P. Tackley. Planforms of self-consistently generated plates in 3D spherical geometry. Geophys. Res. Lett., 35:L19312, doi:10.1029/2008GL035190, 2008.

J. van Hunen and J.-F. Moyen. Archean subduction: Fact or fiction? Annual
Review of Earth and Planetary Sciences, 40(1):195–219, 2012. doi: 10.1146/

J. Walker, P. Hayes, and J. Kasting. A negative feedback mechanism for the long-term
stabilization of Earth’s surface temperature. J. Geophys. Res., 86:9776–9782, 1981.

P. Ward and D. Brownlee. Rare Earth. Copernicus – Springer Verlag, New York, 2000.

J. M. Warren and G. Hirth. Grain size sensitive deformation mechanisms in naturally
deformed peridotites. Earth Planet. Sci. Lett., 248(1-2):438–450, 2006.
doi: DOI:10.1016/j.epsl.2006.06.006.

M. B. Weller and A. Lenardic. Hysteresis in mantle convection: Plate tectonics systems.
Geophysical Research Letters, 39(10), 2012. doi: 10.1029/2012GL051232.

A. Yin. Structural analysis of the Valles Marineris fault zone: Possible evidence for largescale strike-slip faulting on Mars. Lithosphere, 4(4):286–330, 2012.

S. Zhong, M. Gurnis, and L. Moresi. Role of faults, nonlinear rheology, and viscosity
structure in generating plates from instantaneous mantle flow models. J. Geophys. Res.,
103:15255–15268, 1998.