See also Colorado Geology Overview, Colorado Rocks
Last modified 01/16/05
A good understanding of Colorado's physical history and rocks requires a feel for large-scale geodynamics. This groundwork article attempts to provide just that context while supporting the photojournals and other groundwork articles that draw on such ideas.
Most importantly, this article seeks to provide a basis for appreciating if not answering the $64,000 geo-question about Colorado:
As of 2004, no one even pretends to know the answer, but it sure is fun to think about. To many trained minds, anomalous upper mantle stirrings and enduring lithospheric weaknesses must enter the Colorado equation, but how and why remain as big a pair of mysteries as you'll find in geoscience.
Speaking of mysteries, you may wonder why marine processes receive so much attention in this article when the nearest true ocean is over 1,000 km away and has been for at least 1.4 Ga. There are two good reasons. First, much of Colorado's landscape came from the sea, however continental it may seem today. Its basement was largely cobbled together from a host of island arcs and backarc basins between 1.78 and 1.65 Ga, and marine deposits as young as ~70 Ma make up a good bit of its sedimentary cover. Together, these marine contributions account for much of the rock exposed in Colorado today. Second, subduction of ocean floor has driven the planet's surface dynamics since at least the onset of plate tectonics at ~2.0 Ga. Details of the subduction process and of ocean floor creation and anatomy become pertinent at many points in Colorado's story.
Since its emergence in the early 1960s, the theory of plate tectonics has evolved rapidly. Today, it's a theory only in the sense that relativity and quantum mechanics are theories: The testable predictions having been thoroughly confirmed, no one seriously doubts the basic premises at this juncture, but significant challenges and shortfalls remain here and there. As a site of substantial and long-lived intraplate deformation, Colorado embodies more than its share of the remaining difficulties in plate tectonics, but the state's early evolution can only be understood within that framework.
Through the pioneering work of a small group of geophysicists and physical oceanographers, the fundamentals of the theory of plate tectonics
fell quickly into place in the 1960s. Once the fundamentals and their immediate implications were taken seriously, many important details of the theory quickly took shape. Among the founders were Harry Hess, Robert Dietz, J. Tuzo Wilson, Ron Mason, Fred Levine, Drummond Matthews, Lawrence Morley, Walter Pitman, Bill Menard, Bruce Bolt, Jack Oliver, Xavier Le Pinchon and Dan McKenzie, many of whom worked under essential facilitators like Maurice Ewing of Columbia's Lamont-Doherty Geological Observatory and Teddy Bullard of Cambridge.
To a large extent, the early work in plate tectonics was provoked by startling new geophysical observations coming out of
Once people started looking at the ocean floor in earnest, arresting patterns of topography, magnetism (right) and global seismicity quickly emerged. Never before seen or even imagined, these new data sets demanded explanations that theories of the earth based on fixed continents simply could not provide.
The earliest plate tectonic notions were easily summed up in grossly oversimplified cross-sectional cartoons showing plates spreading apart, subducting, colliding, and sliding past each other at homogeneous plate boundaries. The only possible driving force for plate motions anyone could think of at the time was simple pot-on-the-stove Rayleigh-Bénard convection within a homogeneous single-layer mantle.
Plate tectonics version 2.0 developed in the 1970s and 1980s with the recognition that
The simple 2D cross-sections of the early days were no longer adequate, and the role of the mantle had become at once more complex and less clear. Among the version 2.0 pioneers making lasting contributions were Tanya Atwater, Peter Molnar, Warren Hamilton and Bill Dickinson.
As attention turned to the continents and hot spots (excessive volcanic disturbances like Hawaii), the theory stretched to incorporate mantle processes not directly tied to plate motions. In 1971, W. J. Morgan introduced the idea of a narrow plume of heat rooted in the lower mantle to explain the long angled trace of the Hawaii-Emperor chain (above). Plumes were soon invoked to explain all kinds of hot spots, including midplate magmatism at Yellowstone, isolated ocean islands like the Galapagos, oceanic basaltic plateaus like Iceland and triple junctions like Afar. Suddenly plumes were everywhere, often in places that could be explained much more simply without them.
Note that version 1.0 and 2.0 ideas continue to dominate both professional and popular geologic accounts—particularly those in the displays and books offered at tourist attractions of geologic interest.
By the late 1990s, geoscience found itself struggling to move beyond earlier formulations of plate tectonics, just as the founders of plate tectonics had struggled against the fixed-continent establishment of their day. Among the pioneers of plate tectonics version 3.0 are geophysicists Don Anderson and Warren Hamilton. This revision is still in progress.
It's been slow to die, but Morgan's plume model continues to lose credibility as ever more reliable and far-reaching geophysical observations tighten constraints on the workings of the plates and the mantle at depth. As the structure of the mantle and the importance of extensional stresses in locating hot spots became clearer, lower mantle convection, if it occurs at all, had to be decoupled from surface processes, and upper mantle convection had to be seen as a result far more often than a cause of plate interactions.
In plate tectonics, version 3.0, the lithosphere is viewed as a far-from-equilibrium open thermodynamic system of self-organizing semi-rigid plates acting to dissipate crust and mantle heat. The plates are weak in extension and permeable to melt from the underlying, barely solid asthenosphere. Top-down cooling and gravity are now seen as the exclusive driving forces behind plate motions. The largely passive upper mantle receives heat from the lower mantle only by conduction and cools primarily through seafloor spreading (60%) and subduction. Subduction-related hinge rollback and overriding plate extension have emerged as the fundamental shapers of the face of the earth. Hot spots are now seen as excessive volcanism focused by extensional plate failures or, in rare cases, by local and relatively shallow thermal disturbances related to upper mantle temperature variations induced by nearby plate motions.
This article presents the planetary structure and dynamics underpinning Colorado's physical evolution from a plate tectonics version 3.0 perspective. In this endeavor, I owe a great debt to Warren Hamilton, whose gift of an informal but incisive personal education in v.3 tectonics I'll always cherish.
The only thing better than a home planet full of beauty, power and fascination is one that's finally starting to make some sense.
Skip to Crust
Since its initial Early Proterozoic assembly via subduction-related collisions at 1.78-1.65 Ga, Colorado has been far removed from active plate boundaries, but far-field effects from distant plate interactions to the west, south and east continued to play important roles in its evolution well into Tertiary time, both at the surface and at depth. Before launching into the planetary dynamics needed to put Colorado's physical evolution into a current plate tectonic framework, we'll need to spend some time getting to know the playing field and the players.
Having a hard time remembering the Cretaceous? To brush up on geologic time scales and nomenclature, try this deep time refresher. You might also want to keep a browser window open on the ^University of California (Berkeley) Museum of Paleontology Web Geological Time Machine. You'll find links to additional geologic timelines and glossaries on the search page.
The current planetary structure described below had quite literally fallen into place by ~4.4 Ga, but plate tectonics has operated on planet Earth only in the last ~2.0 Ga (from 13:20 hours on). Prior to that, a much hotter earth lacked the crustal and upper mantle strength to support discrete continents standing high above ocean basins, and very different surface processes applied. If you're interested in pre-plate details, take a side trip to the Earth before plate tectonics or a brief thermal history of the Earth.
As every school kid and parent knows by now at some level, the earth (right) consists of
The layers have cooled, and the upper mantle and crust have stiffened up considerably since then, but this basic solid-earth structure has been in place since ~4.4 Ga.
This article focuses on the crust and the upper mantle, which together supply both the players and the playing field for plate tectonics. Via climate, the hydrosphere and atmosphere strongly influence surficial geologic processes like weathering, erosion, sediment transport, deposition and diagenesis, isostatic rebound and subsidence, but they're still passive players at the plate tectonic level. The lower mantle contributes only gravity and some heat to the process.
Largely isolated as it is from surface processes by the profound thermodynamic and compositional barrier at 660 km, we can safely ignore the workings of the lower mantle here. And since the core contributes little to our story beyond gravity and the geomagnetic field, we'll ignore it entirely.
In addition to the familiar compositional layering outlined above, it's essential to recognize an overlapping mechanical subdivision of the planet at depths above 660 km — into lithosphere, asthenosphere and remaining upper mantle. These overlapping boundaries make it hard to understand what "upper mantle" means in some contexts, but I'll try to be as clear about that as I can without adding too many extra words. Geoscience could really use a short recognized term for "remaining upper mantle", or what the table at right calls "lower upper mantle".
Since ~2.0 Ga, the crust and the cool, stiff layer of mantle immediately underlying it have moved together over the surface of the earth as a segmented but fairly coherent planetary shell called the lithosphere. By definition, the layer of mantle that tags along with the crust is the lithospheric mantle.
The plates of plate tectonics are nothing more than mobile lithospheric fragments — some large, some small, some mostly oceanic, some mostly continental, but all at least oceanic in part. The plates slide around on a slippery, barely-solid, ready-to-melt layer of upper mantle known as the asthenosphere that leaks out to the surface whenever it can. Like it or not, we're along for the ride
About 85% of the earth's surface can be understood in terms of physical interactions among plates and between plates and the asthenosphere. That is the fundamental lesson of plate tectonics. Colorado falls squarely in the remaining 15%, where other processes, perhaps of mantle origin, also seem to come into play.
The earth is a mighty heat engine constrained by its own gravity. To understand its workings, one must first sort out how the temperature, pressure and the density of its materials vary with depth.
Temperature and pressure increase steadily with depth throughout the earth but along different curves and more steeply at some levels and locations than at others.
The geothermal gradient, the curve describing the change of temperature with depth, reflects the cooling of the earth. It's fairly well known at most depths from well logs and geophysical observations, including the behavior of planet-crossing seismic waves. The global average is ~25°C/km, but the observed range is 5-90°C/km. At shallower depths, the geothermal gradient is flatter and varies considerably from place to place. For instance, at 10 km down, the temperature is 300°C under Los Angeles but only 150°C under Pittsburgh.
Pressure gradients in the earth are also fairly well worked out, at least at some depths. At all depths, the pressure is largely lithostatic—i.e., simply due to the combined weight of all overlying materials. In the crust and upper mantle, plate tectonic processes can vary pressure a bit around lithostatic values—not by much, but enough to make a difference, particularly vis-à-vis the asthenosphere's very tenuous hold on solidity. The unthinkable lithostatic pressures found at lower mantle depths severely suppress convection but foster chemical stratification.
For the most part, density also increases steadily with depth, particularly at the core. There is, however, one all-important exception:
The earth has been losing heat to space since it first formed but remains very hot internally, even after 4.5 Ga of cooling. Major impacts during the Hadean Eon (4.5-3.8 Ga, 00:00-03:40 hours) provided significant setbacks early on, but the planet's been cooling steadily ever since. The heat the earth now sheds into space comes from several non-renewable sources:
To put the importance of radiogenic heating into perspective, note that the heat flux at the surface is ~10 times that across the core-mantle boundary. Note further that radiogenic heat generation now is probably 2-3 times less than it was in the Archean Era (3.8-2.0 Ga) due to intervening radioactive decay.
Why bother with the cooling of the earth? Because processes in the mantle and crust have evolved both quantitatively and qualitatively as temperatures have fallen. Many high-temperature Archean processes have never occurred since, including the formation of continental cratons and their buoyant keels, the formation of domal granite-and-greenstone complexes, the intrusion of large anorthosite bodies, and the eruption of ultramafic komatiite lavas. These events required crust and upper mantle temperatures much higher than today's.
Since ~2.0 Ga, most of the heat escaping the earth does so through the oceans, primarily via seafloor spreading (60%) and subduction. These processes became possible only when the crust and uppermost mantle had become cool and stiff enough to support continents standing high above ocean basins, and the upper mantle had cooled enough to allow the gabbro-to-eclogite phase transition to occur at a shallow enough depth (now ~60 km) to foster subduction. Without this critical transition, which renders falling slabs of oceanic crust denser even than the upper mantle below the asthenosphere, gravity could not effectively drive subduction, and subduction could not effectively drive plate tectonics.
The lower mantle has probably changed little since its differentiation at ~4.4 Ga. Any residual temperature or density contrasts left in the lower mantle at that time were effectively frozen into place. It's circulation is very slow on time scales comparable to the current age of the earth, never mind the time frames within which surface processes evolve.
Skip to Plates and Mantle
Let's start our detailed dissection of the earth with its most tangible layer, the crust. The crust is the surface layer we inhabit and admire. It sits between the atmosphere or hydrosphere on one hand and the rarely exposed mantle on the other.
The accepted boundary between the crust and mantle is the Mohorovicic discontinuity, or Moho for short — a seismographic feature defined primarily by an abrupt increase in the seismic wave velocity (~7 km/sec above and ~8 km/sec below). Away from adornments like seamounts and submarine basaltic plateaus, the Moho on oceanic plates lies a fairly constant 5-6 km below the surface of the crust. On continental plates, the depth to the Moho is all over the lot, with a global mean of ~30 km and a 5-100 km range. On the continents, the Moho also coincides with related contrasts in consistency (stiffer below despite higher temperatures) and in the abundance of high-pressure minerals (more olivine, pyroxenes and garnet below).
On this planet, crust comes in four basic flavors— continental, oceanic, arc-generated, and collisional. Each type forms in a very different way, with resulting differences in composition, structure and mechanical properties.
Geologically, it's useful to divide the crust, whether continental or oceanic, into basement and cover layers. In most places, a stack of relatively thin, weak and easily-eroded sedimentary or volcanic layers (the cover) blankets a much thicker substrate of hard, resistant crystalline (igneous or metamorphic) rock (the basement) extending down to the brittle-ductile transition if not to the Moho. The basement is the foundation of the upper crust and the source of its mechanical strength. It's generally fairly homogeneous and isotropic, meaning that applied stresses find no particular directions or planes of weakness within it away from identifiable faults and shear zones.
The shape of Red Canyon (right) in the Colorado National Monument shows how very differently cover and basement tend to erode. Red Creek cut a wide and deep swath through the sedimentary cover (Wingate Sandstone) at Red Canyon. The meager notch in the distance is the best it could do in the hard 1.7 Ga Precambrian basement flooring the canyon.
Why even mention oceanic crust in a web site about Colorado geology? For starters, many of the exposed 1.7 Ga metamorphic gneisses and schists of the Colorado basement derive from oceanic crustal elements caught up in the collisional crust formed by the many ocean and backarc basin closures involved in the Early Proterozoic Colorado Orogeny. Some have been altered beyond recognition by heat and pressure attending deep burial during collisions and subsequent intrusions, but many retain signatures of their oceanic ancestry—as we see, for example, in the 1.7 Ga metasediments and metavolcanics exposed around the Mt. Evans batholith, in the east wall of Tenmile Canyon, and in Painted Wall of the Black Canyon of the Gunnison (right).
The plain crust flooring the abyssal plains of the oceans is the simplest case. This predictable, mostly igneous 4-layer package is usually ~6 km thick.
Layers 1-3 make up oceanic crust's stable basement, which rests in turn on lithospheric mantle. By mass, the gabbro predominates. Volcanic appendages like seamounts, submarine basaltic plateaus and aseismic ridges can thicken Layer 1 by several kilometers.
On average, oceanic crust is 3-10 times thinner than continental crust. A more mantle-like average mineralogy also makes it much denser (~3,000 vs. ~2,700 kg/m3). High magnesium and iron and low silicon and aluminum contents earn oceanic crust the compositional designation mafic, after the chemical symbols for magnesium (Ma) and iron (Fe). On the whole, oceanic crust much more mafic than the continental crust but less mafic than the ultramafic mantle.
The stable oceanic basement forms simultaneously with its underlying lithospheric mantle at seafloor spreading centers like the Juan de Fuca ridge at right. Together, these two layer make an oceanic plate or slab. The slab's mantle layer thickens over time as chilled asthenosphere freezes to its undersurface in a process called underplating, but the basement changes little until the plate subducts.
Sooner or later, all oceanic crust subducts. The oldest dated continental crust on earth contains ~4.2 Ga zircons, whereas the oldest crust on the ocean floor today is only ~180 Ma old. The average surface dwell time for oceanic crust is a mere ~100 Ma. Satisfactorily explaining these vast age differences was an early plate tectonic triumph.
Varying amounts of Layer 0-1 sediments and basalts get scraped into accretionary wedges during subduction, but much of the sediment goes down with the ship. Occasionally, more complete scraps of oceanic crust called ophiolites manage to escape subduction via subduction zone collisions, which can also elevate thick Layer 0 sedimentary packages like those found on outer continental shelves and slopes to great heights, as in the Andes at left. For the most part, however, subducted oceanic crust just falls back into the upper mantle along with its underlying lithospheric mantle. At a depth of ~60 km, the basement's basalt and gabbro undergo a phase change to eclogite, a metamorphic rock of even greater density. The density kick then accelerates the plate's fall like a sinker on a fishing line.
The easily magnetized oceanic crust functions much like the magnetic coating on a recording tape. (In this analogy, the lithospheric mantle is the tape's plastic base.) Magnetic orientations frozen into cooling mafic basement minerals at the spreading center record the plate's time and place of birth relative to the prevailing geomagnetic field, which reverses at irregular intervals of ~0.5-10 Ma for poorly understood reasons. The field reversals create striking magnetic stripes in the oceanic crust, as diagrammed at right.
Geoscientists have made great use of the remanent magnetization of the ocean floor by cross-referencing its patterns against a hard-won timetable of geomagnetic field reversals. In fact, magnetic mapping of the sea floor first cracked open the plate tectonics revolution in earth science back in the early 1970s.
Freshly formed basalt-covered oceanic basement is bare. It acquires a sedimentary cover as it rafts away from its spreading center of origin. The sediments provide a variably detailed record of the travels and encounters of the plate on which they amass. In well-developed, sediment-starved deep ocean basins like the Atlantic, the sedimentary veneer rarely exceeds a few hundred meters in thickness. Uninterrupted accumulations of pelagic (deep ocean) sediments like radiolarian chert record time spent on the abyssal plain, far from land and well below the carbonate compensation depth, where sinking calcareous marine animal skeletons (those made of calcium carbonate) dissolve back into sea water, usually ~4 km down.
Terrigenous (land-derived) sediments on oceanic crust tell of formation or passage near an eroding landmass. Continental shelves are thick sedimentary sequences piled for the most part onto faulted continental crust at the passive margin of a successful continental rift, but they sometimes advance over oceanic crust, as they do along the margins of the Red Sea. Volcaniclastic sediments resting on oceanic crust identify the nearby landmass as a magmatic arc. Large, mature oceanic arcs like Sumatra (right) can shed substantial shelf-like sedimentary aprons onto oceanic crust in both forearc and backarc settings.
Turbidity currents are the prime movers of terrigenous sediments onto oceanic crust. They're also the prime carvers of the submarine canyons found in river deltas, continental shelves and other sedimentary aprons. Turbidity currents form when coastal waters heavy with silt periodically roil down submarine canyons or along trenches to spread their loads onto the deep ocean floor in large volumes, sometimes over distances of hundreds of kilometers. For example, turbidity currents coming off the Ganges-Brahmaputra delta at the north end of the Bay of Bengal carry sediments along the Sunda trench as far south as Bali. The resulting distinctive sedimentary rocks, called turbidites, contribute heavily to accretionary wedges at subduction zones and also to continental slopes, particularly around the mouths of major rivers.
Hydrated low-density sialic (silicon- and aluminum-rich) crustal residues floating on a global ocean of dense hydrated mafic (magnesium- and iron-rich) magma probably developed shortly after the time of earth's formation at 4.5 Ga and may have persisted well into the Archean Eon. Buoyant patches of sialic crust eventually coalesced into a number of large, durable, unsinkable rafts that became the cratons (stable cores) of our modern continents — e.g., the Wyoming Province and the Canadian shield. Through fractionation and underplating of subjacent mantle, these early Archean cratons developed thick (>100 km) hulls and very deep (>250 km) keels of cool, strong, buoyant and refractory lithospheric mantle that to this day protect them from disruption by heat and currents in the underlying upper mantle.
Most of the planet's supply of sialic continental crust formed in the hot early Archean via high-temperature processes never repeated since. Much of that crust was recycled into the upper mantle prior to the advent of plate tectonics at ~2.0 Ga. The remaining Archean cratons have been tied up in continental cores ever since. Continental crust formed since 2.0 Ga, primarily at post-Archean collisional and convergent plate boundaries, has been much more mantle-like (mafic) than the surviving Archean cratons.
BTW, Colorado contains no cratonic crust to speak of, but adjacent Wyoming is a different story.
Relative to oceanic crust, the crust found in cratons is usually much lighter (~2,700 vs. 3,000+ kg/m3), much thicker (for the most part, 20-60 km vs. ~ 6 km) and much more complex with regard to both structure and composition. Continental crust is typically composed of crystalline igneous and metamorphic rock at depth and all manner of sedimentary, igneous and metamorphic rocks near the surface. On average, sialic continental crust contains much more silicon and aluminum and much less magnesium and iron than the mafic oceanic crust and the ultramafic mantle. Rocks of this composition are said to be felsic, because they contain abundant feldspars (K, Na, Ca, Ba, Rb, Sr and Fe aluminum-silicon oxides) and silica (Si02).
Mechanically, continental crust divides into a cooler, stronger, brittle upper crust and a hotter, weaker, ductile (putty-like) lower crust. An important thermodynamic and mechanical boundary known as the brittle-ductile transition (BDT) separates the upper and lower crust, usually 10-15 km below the surface. The terms middle crust and mid-crust usually refer to a horizon around the brittle-ductile transition.
The hot, plastic ductile lower crust flows, slowly of course, under the influence of gravity and regional stress fields, just as an incandescent but quite solid steel ingot flows to take on the shape of a I-beam under the force of a hydraulic roller press. The less compliant brittle upper crust folds, buckles and snaps when stressed beyond its limits but is nevertheless the main locus of mechanical strength in continental lithosphere. The upper crust generates most of the world's earthquakes, but the lower crust can get into the act now and then, too. Below the Moho, the ductile mantle is itself seismically silent. All subcrustal earthquakes originate in still-brittle slabs of oceanic crust falling through the upper mantle below subduction zones but above the "660", the effectively impenetrable physicochemical boundary between the upper and lower mantle. Claims of earthquakes (and therefore brittle behavior) below the "660" are based on highly suspect data.
Over most of the globe, the upper crust further divides into cover and basement layers, each with markedly different rock types and mechanical properties. The strength of the upper crust and to a large extent, that of the entire continental lithosphere, resides in the basement. The cover may have its own secondary BDT, but it's the primary basement BDT that separates the upper from the lower crust.
We'll examine this important intermediate form of crust in detail when we discuss the magmatic arcs that accompany subduction below. Suffice it to say here that arc-generated crust is complex and highly variable but on average falls somewhere between mafic and felsic in composition. The Colorado basement includes a substantial portion of metamorphic arc-generated crust of Early Proterozoic age. At right is the Costa Rican magmatic arc built on the southwest margin of the Caribbean plate, where the Cocos and Nazca plates subduct along the dotted cyan line.
Continents add mixed crustal materials at or near their margins through subduction zone collisions of all kinds. Over deep time, long-lived continental arcs can accrete substantial volumes of arc-generated crust of intermediate mafic-to-felsic composition, fragments of partially subducted mafic aseismic ridges, submarine basalt plateaus and ocean islands, slivers of mafic oceanic crust and occasional felsic continental fragments (e.g., ribbon continents) as they scrape or break these buoyant incoming structures off falling slabs. Between mid-Paleozoic and Late Jurassic time, North America grew by a third or more along its western margin through the accretion of many such far-flung exotic terranes, some of which traveled north 3,000 km or more to become California, Nevada and parts of Alaska and Utah.
Lesser collisions at continental arcs many not result in terrane accretion, but they can still generate complex crust by crushing backarc basins and arcs together. Submarine basaltic plateaus and aseismic ridges can subduct successfully if not gracefully. Their buoyancy can generate enough compressive stress in the overriding plate to close and invert a backarc basin like the Sea of Japan (right) and can even mix the arc itself back into the resulting orogen. A new arc may then form outboard of or sometimes over the mess created by the collision. In fact, backarc basins can form and close repeatedly, accordion-like, along the same continental margin in a process known as tectonic switching. Backarc basins are important nurseries for collisional continental crust.
Continent-continent collisions may also entrap exotic terranes other oceanic elements as they close both ocean and backarc basins in complex time-progressive patterns along strike.
However they form, these mixed-bag continental add-ons, called mobile or orogenic belts, can produce extraordinarily thick, complex and puzzling crustal structures and juxtapositions overprinted by metamorphism varying widely in both degree and P/T regime. Things get even crazier when erosion and later deformations and magmatic events begin to tamper with the evidence. Between 1.78 and 1.65 Ga, the Colorado Province quickly grew into a large and very complex mobile belt through backarc mashing and the accretion of many primarily oceanic terranes. Nearly 1.7 Ga later, after long bouts of deep erosion and several major superimposed deformations and intrusions, most of its pieces are hopelessly scrambled into the basement, and some are missing or metamorphosed beyond recognition. Nevertheless, basement exposures like those in sections of the Black Canyon of the Gunnison occasionally preserve enough original structure and composition to give an idea of what went on.
To get an even better feel for what students of Colorado's Precambrian past are up against, take a look at the modern Banda Sea (right), which contains many oceanic elements and a few continental fragments as well. As Australia advances on Southeast Asia, the sea will close, probably in the next 10 Ma or so. Geologists studying the resulting suture 50 Ma from now will have a heck of a time unraveling all the pieces, even without further tectonism. I wish them luck.
Mobile (orogenic) belts are less stable than the cratons they encrust for a number of reasons. Structurally, they're usually on thinner lithosphere, even when they include thickened crust, because they lack the deep refractory keels that insulate cratons from potentially disruptive upper mantle heat and currents. As a natural consequence of their creation in collision zones, they tend to be hotter and more densely and thoroughly faulted, with more built-in zones of weakness of lithospheric scale. Their greater compositional diversity can include weaker materials, and on average, they're more mafic and therefore less buoyant relative to the asthenosphere. As a result, they're more likely than cratonic crust to pass from continent to continent in the supercontinent cycle, or to disappear down a continent-continent subduction zone, temporarily or permanently. That's why they're called mobile.
Nearly all of the planet's oceanic crust gets recycled through the upper mantle via the processes of subduction and seafloor spreading. Oceans as we know them have been around since at least ~2.0 Ga and probably in much shallower form since at least 3.5 Ga, but nowhere is today's oceanic crust older than ~180 Ma. With the exception of a few slivers (ophiolites) squeezed into mobile belts here and there, all older oceanic crust has been returned to the upper mantle by subduction. The average surface dwell time for oceanic crust is a mere ~100 Ma.
Continental, collisional and arc-generated crusts are a much different story. They're too buoyant to subduct in any significant quantity (try pushing a basketball underwater), so they're usually left behind at the surface at subduction zones, where they generate more collisional crust. They get passed back and forth among the continents du jour as the plates do their dance in the supercontinent cycle. Continental mobile belts are particularly susceptible to exchange among the seldom-rifted continental cores of ancient cratonic crust, but the cores change partners among themselves. The oldest exposed continental crust, part of such a core in southwest Australia, dates back to 4.2 Ga.
The continents slowly lose crustal mass "off the top" to minor subduction of the marine sediments they shed, but most of their eroded materials are simply redistributed and recycled into different continental crust via processes ranging from sediment transport and deposition to the formation of collisional crust to the supercontinent cycle. Total continental area has grown slowly since the inception of plate tectonics at ~2.0 Ga through the formation of mobile belts, primarily by way of arc magmatism and backarc spreading.
Skip to Plate Motions
The earth's outermost shell, the lithosphere, is segmented into semi-rigid mobile plates that give the planet its facial expression. The plates are fairly rigid internally, particularly around the ancient continental cores known as cratons, but are variably deformable at their margins. At times, plates also deform far inboard, as the Rockies and the Colorado Plateau loudly testify.
In aggregate, the plates form a self-organizing, far-from-equilibrium dissipative system driven only by gravity and the top-down cooling of the earth, subject to spontaneous reorganization without notice on long and short time scales and on large and small distance scales. In other words, the plates behave chaotically in the mathematical sense of the term. The upper mantle moves primarily in response to the motions of the plates, not the other way around, as originally thought. Interestingly, there always seem to be about 12 roughly pentagonal major plates, with many minor plates filling in the gaps.
Plates (like the Pacific) that include subducting oceanic lithosphere are the freest to roam, and they contribute most to the changing face of the earth over time. Most plates lack subducting boundaries, however. Their wanderings are heavily restricted by the motions of adjacent plates. Together, the plates form a global circuit of mutually related motions that can get pretty complicated at times.
You can watch the plates dance to the beat of gravity through time in the animations at the ^PALEOMAP Project. You can examine the plates and their current relationships in many fascinating ways at the Jules Verne Voyager, Jr. interactive mapping site, where I produced the map at right.
Plates carry their crust on a relatively cool, stiff basal layer of outermost solid mantle called the lithospheric mantle, simply defined as the portion of the upper mantle that moves in lockstep with the crust. Together, the crust and lithospheric mantle make up the lithosphere. The table below may help you sort out these overlapping divisions of the earth.
The lithospheric mantle slides around on a very hot, barely solid, highly deformable upper mantle layer known as the asthenosphere, from the Greek for "weak shell". Extending down to a depth of ~200 km, the unstable asthenosphere is ready to melt at the slightest provocation. More often than not, the excuse is relief of lithostatic pressure (depressurization) secondary to extension within or between plates, as at spreading centers and subduction zones. The strong global concentration of volcanoes along plate boundaries is no accident.
Compared to the stiff, groaning crust, the lithospheric mantle is weak and seismically silent, and the asthenosphere is like so much grease under the skids.
Mechanically speaking, a plate's strength (i.e., its ability to resist deformation and fracture under tensile, compressional and bending stresses) resides in its stiff upper crust—the only portion brittle enough to generate earthquakes. The weaker, ductile lower crust and lithospheric mantle (slightly stiffer and denser than the lower crust but still weak relative to the upper crust) are supported by the hard shell of upper crust in exactly the same way that a crab's exoskeleton supports its soft internals. The hot asthenosphere is too plastic to support much of anything except by buoyancy, but since all continental lithosphere literally floats on the asthenosphere, that's no small contribution.
Plates support large mountain ranges in 2 main ways—through flexural strength and through buoyancy. The diving board's flexural strength supports the diver at its tip as she gathers herself, eyes closed and motionless. In the same manner, the flexural strength of the thick upper crust of the partially subducted northern margin of the Indian plate supports the soaring height and immense weight of the Himalayas and the Tibetan Plateau (top center at right), both of which stand on the shattered and weak overriding southern margin of the Eurasian plate. Without the flexural support of the Indian plate, the Himalayas would sink and drown in their own sediments.
The majority of mountain ranges not associated with continent-continent collisions receive little flexural support from surrounding crust. Instead, such ranges support their locally thickened crust through buoyancy relative to denser rock below. Once you take to heart the viscous behavior of rocks below the upper crust, this is nothing more than a straightforward application of Archimedes principle. Isostasy is a fancy name for the obvious requirement that
Otherwise, lithostatic pressure differences would force denser, more ductile rock below the upper crust to flow into underweight columns until the pressures equalized.
Since earth materials are too weak and earth processes too slow to generate dynamic (tectonic) overpressures large enough to override isostasy, it follows that equally large roots of light-weight crust should underlie and buoy up all mountain ranges lacking flexural support. Most ranges not associated with continent-continent collisions have geophysically detectable roots within the crust—among them the Andes (right) and the Sierra Nevada but not the Rockies.
Note: To understand isostasy, think of a coal barge. The higher you heap the coal, which is less dense than water in aggregate, the lower the barge sits in the water. The submerged portion is the root. As long as the barge floats freely, it has only isostatic (buoyant) support. If the barge were tightly cabled into a large raft of barges, it would then receive some flexural support from adjoining barges via the cables. If it were pulled across the water at high speed, it would also receive dynamic support via inertial forces exerted by the water.
The Rockies seem to be a very peculiar case here. They now stand at great average elevation over 1,000 km from any plate margin without any crustal roots or flexural support to speak of, and have done so in the face of erosion for perhaps some 70 Ma—quite a long time as mountain ranges go. In fact, they're as high as ever and still rising!
What keeps them up then? The best bet is isostasy operating at upper mantle rather than at the usual crustal levels. A large low-velocity upper mantle seismic anomaly (the Aspen anomaly) underlying the central Rockies is widely interpreted as a zone of higher than usual temperature. What that means is far from clear, but if lower seismic velocities also mean lower densities, as they should at the depths involved, then the upper mantle may be providing deeper than usual buoyant support for the Rockies. Dynamic support by a regional upwelling of the upper mantle (an inertial "jet" effect) is unlikely. Intraplate disturbances have repeatedly rumpled this portion of the continent for at least 1.4 Ga, since the time of the Berthoud Orogeny. That kind of time frame also raises the question of a long-standing lower mantle influence. More at 11...
Whether continental or oceanic, plates have little tensile strength. In other words, they resist compression fairly well but pull apart easily. Most if not all large volcanic outbursts (hot spots) result, not from mantle outbursts and thermal anomalies, but from the failures of plates in tension. Indeed, the plate need not crack overtly; all that's required is a horizontal axis of least compressional stress. (This is known as the diking condition.) Once it gets a foot in the door, the buoyant force behind a depressurization melt does most of the work involved in breaching the plate. A single meter-wide dike can spew out tremendous volumes of lava in geologically short timeframes. In all likelihood, the intense and widespread pulse of magmatism that swept Colorado in the mid-Tertiary resulted, at least in part, from regional extension associated with the passing of the Rio Grande Rift.
As your read on, keep this firmly in mind:
Spreading centers like the Mid-Atlantic Ridge and the East Pacific Rise; ocean islands like ^Iceland and the ^Emperor-Hawaiian chain; flood basalts like the Snake River Plain and the ^Deccan Traps of India; leaky faults like the San Andreas transform and the many normal faults of the Basin and Range; massive intraplate volcanic disturbances like ^Yellowstone; and vast oceanic plateaus like the Ontong Java all owe their existence to this fundamental fact of life on planet Earth. What comes out depends critically on how thoroughly the asthenosphere melts and on what the melt passes through to reach the surface, but basalt is the quintessential leakage lava, especially in oceanic settings.
The lithospheric mantle and the asthenosphere are the uppermost layers of the upper mantle, which ends at depth at the "660"—the fundamental upper-lower mantle boundary found globally at 660±30 km. An obvious discontinuity in seismic probes of the mantle, the 660 appears to represent a solid-solid phase transition localized at 660 km by relatively fixed temperature and pressure gradients.
In the diagram at right, the 660 runs along the top of the yellow lower mantle. The more complex upper mantle consists of a basal layer (salmon), the mobile asthenosphere (orange) and the plate-bound lithospheric mantle (gray).
Due to a number of mutually-reinforcing physicochemical influences, material has a very hard time crossing the 660 in either direction, if it crosses at all. This barrier has profound implications for earth dynamics. For starters, it means that
Subducted oceanic lithosphere drops through the upper mantle and piles up along the 660 to await recycling. If the the lower mantle convects at all, it does so very slowly in isolation below the 660, passing heat to the upper mantle only through conduction. Observed and inferred currents in the upper mantle derive primarily from plate motions and from small-scale convection cells related to plate boundaries, not from currents in the lower mantle.
Thanks to the self-organizing plates, the upper mantle is heterogeneous in both temperature and composition. It has absorbed a great deal of oceanic lithosphere since the onset of plate tectonics at ~2.0 Ga, and it absorbed a good deal of felsic Archean crust before that. As a result, it remains rich in volatile elements (C, H and O) and also in fertile minerals (those with low melting points). Among other things, these constituents make the asthenosphere—a key player in both plate tectonics and intraplate volcanism—all the more eager to melt and the deeper upper mantle less dense than sinking slabs of oceanic crust. We'll return to the upper mantle many times as this overview of earth dynamics unfolds.
By far, the lower mantle's greatest contribution to observable geologic processes is the gravity generated by its immense mass—well over half that of the planet. Since gravity is main drive in most surface processes, including plate tectonics, gravitational collapse, mass wasting, erosion and sediment transport, that's no small matter, but the lower mantle's influence pretty much ends there.
The lower mantle appears to be stratified chemically, with likely boundaries at 1,000 and 2,000 km. If it convects, it may do so very slowly in several layers isolated from each other and also from the upper mantle and crust. Having lost most of its heat-generating radioactivity to the crust and upper mantle in the planet's thorough chemical differentiation prior to ~4.4 Ga, for most of earth's history, it has passively conducted original heat amounting to only ~10% of the surface flux from the core to the upper mantle with little help from convection.
Can lower mantle features have surface expressions? Theoretically, density lows in the lower mantle should lead to broad topographic highs at the the surface, while temperature highs should heat the overlying upper mantle with potential effects on plate distribution. At lower mantle densities, viscosities and thermal expansion rates, any such contrasts should be very large and semi-permanent. No such associations have been documented, but I wonder about the curiously rootless Rockies, which lack a demonstrable means of support.
Narrow-based "plumes" of hot material rising directly from the core-mantle boundary to manifest as surface hot spots are physically impossible, despite many educational cartoons and professional publications to the contrary. These plume proponents ignore the convection-stifling effects of pressure on thermal expansivity, viscosity and thermal conductivity at planetary scale. Add the pressure back into the models, and plumes don't form. Nor have any such plumes been detected by geophysical means.
Since the lower mantle has nothing else to do with plate tectonics, and certainly nothing to do with mantle plumes or hotspots, we'll ignore it henceforth.
Since 2.0 Ga, the plates have clumped and split and regrouped many times, but by most counts, there are currently 20 plates—12 major and 8 minor. The 7 largest plates now account for 94% of the earth's surface, oceans included, and the smallest of the 12 major plates is larger than all of the minor plates combined.
The observation that there always seem to be ~12 major plates in the geologic record suggests that a dodecahedral division of the lithosphere involving 12 roughly pentagonal plates may be energetically and statistically favored over time. Experimentally and mathematically, there are only so many good ways to crack up and move a spherical shell held together by gravitationally-induced lateral compression, which acts something like surface tension. In the case of the earth, the weight of the plates and their willingness to spread under gravity at free margins (AKA subduction zones) together provide the lateral compression that keeps plates in contact.
Skip to Plate Boundaries
To understand why plates move, consider these geometric and physical constraints on plate tectonics:
Oceanic plate segments move toward their subduction zones because they are under constant compression, both by their own weight and by neighboring plate segments, and simply have no place else to go. Overriding plates at such subduction zones move or stretch toward the subducting plates as the latter fall out of their way. Through these simple local interactions, the plates enter a ^self-organized dance in the same way that shoals of fish manage to dart in unison without a leader.
You can watch the plates dance through time in the online animations at the ^PALEOMAP Project.
The earth's daily rotation appears to energize, or at least foster, plate motions to some extent. On average, relative plate velocities near the equator are double those near the poles. This stable pattern and the net westerly drift of lithosphere apparent in the Antarctic frame of reference together strongly imply an input from the planet's spin.
Slow convective circulation of hot, plastic mantle rock is often said to be the energy source powering the dance of the plates, but many lines of geophysical evidence rule that out. Upper mantle convection certainly occurs in small scattered cells, but it doesn't exert enough horizontal force on the lithosphere to push plates around. In fact, upper mantle currents result from plate motions, not the other way around. Lower convection, if it exists, is confined below the "660" and is thus effectively decoupled from the upper mantle and lithosphere. Heat rising from the lower mantle enters the upper by conduction, not by convection.
The bottom line:
Relative plate motions are the most important because they control plate interactions directly. Absolute plate motions are also of interest, but they're difficult to pin down for lack of a suitable reference frame. Of the many possible frames proposed, the most commonly used are the hot spot frame pegged to Hawaii or to a collection of selected "fixed" hot spots (on the assumption that the proposed underlying mantle plumes must be fixed relative to the mantle), and the "no net rotation" (NNR) frame designed to minimize all plate motions. But hot spots turn out to migrate at rates (3-6 mm/year) comparable to plate velocities, and the NNR frames assign plate velocities and relative motions that just don't jibe with observed plate boundary types. In other words, plate velocities don't necessarily point toward convergent boundaries, away from divergent boundaries or along transform boundaries, as one would expect.
A simpler and more realistic frame plots recent plate motions against a fixed Antarctica (right) since, more than any other plate, it has boundaries that are nearly all divergent. (The only exception involves the tiny Scotia plate east of Cape Horn.) In the Antarctic frame, all slab hinges roll back (as they must), all ridges migrate (to tap fresh asthenosphere) and plate velocities actually point toward convergent plate boundaries, away from divergent boundaries and along transform boundaries, as they should. You can give the Antarctic and other frames a try online at the Jules Verne Voyager, Jr. interactive mapping site.
For the most part, the plates move as rigid bodies over the sphere of the earth along fairly sharp boundaries and interact and deform primarily along those boundaries. But significant areas of intraplate deformation affect 15% of the earth's surface, including much of Asia within and around the Tibetan Plateau (right).
thanks in part to a 1.6 Ga-long legacy of weakness in the Colorado basement. Mantle stirrings probably also play a role in the Rockies, but relevant data is hard to come by. Intraplate deformations remain an active and challenging area of research in contemporary geoscience.
Skip to Lithosphere
The plates are semi-rigid spherical caps with varying shapes and sizes at the earth's surface. On a planetary scale, most plate boundaries are surprisingly sharp, but some are diffuse—in some cases, up to hundreds of km across. The boundary around any one plate may vary in character from place to place, but the nature of one plate's boundary with any other single plate is in large measure dictated by the relative motion of the two plates involved. Plate boundaries are therefore predominantly one of three basic types—divergent, convergent or transform.
Currently, 80% of the earth's plate boundaries are divergent mid-ocean ridges and convergent subduction zones. As the globe at right shows, most boundaries are on the ocean floors, and this is no accident: By weakening rock and promoting melting, water tends to facilitate plate interactions.
Plates move away from each other at divergent boundaries, most of which are seafloor spreading centers. Mid-ocean ridges like the Mid-Atlantic ridge shown at right are the purest and most developed examples of divergent plate boundaries, but less oceanic examples include the Red Sea and the Gulf of California. Divergent boundaries create all of the planet's oceanic lithosphere.
Plates move toward each other at convergent boundaries. Subduction zones and collision belts mark convergent boundaries, depending on the lithospheric types involved. Most convergent boundaries involve oceanic lithosphere on at least one side, but continent-continent convergences make for some of the planet's most dramatic scenery, including the Himalayas (right) and the Alps.
Plates slide past each other horizontally along transform boundaries, or transforms for short. Over time, transforms have been responsible for many large-scale redistributions of continental crust. California's San Andreas fault is the main locus of motion (5 mm/yr) between the Pacific and North American plates, but their rather diffuse transform boundary also includes faults far to the east beyond the Sierras. At right, the main trace of the San Andreas fault rumples up the Temblor Range across central California's Carrizo Plain.
Things can get pretty messy after 2.0 Ga of semi-rigid plates bumping and grinding helter-skelter on the surface of a sphere. Mixed plate boundaries tend to occur where the margins and relative motions of the two plates involved fail to align neatly.
In California, the Pacific-North American boundary is both diffuse and mixed. In its drive to the NW relative to the California margin of the North American plate, the Pacific plate must also pull away to the west as it slides along the San Andreas fault, which trends too much to the north to accommodate the relative plate motions cleanly. Since plates lack the cohesion to allow gaps to open up between them, ongoing Basin and Range extension carried by gravity-driven lower crustal flow accommodates the westward component of divergence along the predominantly transform Pacific-North American boundary.
So far, we've considered only boundaries involving two plates. The coming together of three plates is called a triple junction, the ridge-ridge-ridge (RRR) variety being the most common. Eventually, all plate boundaries end at triple junctions, as do all seams on a soccer ball. California's Cape Mendocino (right) marks a seismically dangerous trench-transform-transform triple junction terminating the Gorda Trench, the San Andreas Fault, the Mendocino Fracture Zone.
Most RRR triple junctions are oceanic, but the Afar triple junction of Ethiopia and Somalia is a notable exception. Particularly leaky oceanic RRR triple junctions are the most common source for the seamounts and large submarine basaltic plateaus that dot the ocean basins.
Triple junctions exhibit many other surprising behaviors, but they're not a significant part of Colorado's story and won't be discussed further here.
Note: Quadruple and higher-order junctions don't seem to occur and wouldn't be expected for a number of basic physical and geometric reasons related to the way things tend to play out on spherical surfaces.
Skip to It's All About Subduction
The earth's mobile lithosphere consists of the crust and the chilled topmost layer of the upper mantle. These two layers are usually inseparable, but the lithospheric mantle can occasionally peel off (delaminate), as it is doing now beneath the Carpathian Mountains of the Southern Alps. The plates of plate tectonics are lithospheric rafts that glide across the surface of the planet on the barely solid, plastic asthenosphere.
Like crust, lithosphere comes in two basic end-member flavors, continental and oceanic, but intermediate forms occur, particularly at passive continental margins. Most plates include both continental and oceanic lithospheric components.
This is the fundamental fact of plate tectonics.
Geoscientists call oceanic segments of lithospheric plates slabs for short, and I'll do the same. Slabs are created at seafloor spreading centers and consumed at subduction zones. Their materials are continually recycled through the upper mantle in the process. Compared to continental lithosphere, slabs are thin—on average 80-90 km at subduction zones and closer to 6 km when they first form. They are also denser and more flexible. The oldest dated continental crust (and presumably lithosphere) contains ~4.2 Ga zircons, whereas the oldest oceanic lithosphere is only ~180 Ma old.
More importantly, slabs are both denser and stiffer than the hot, barely solid asthenosphere on which they temporarily rests. This density inversion drives subduction and all of plate tectonics in turn.
Many processes create continental lithosphere, but oceanic lithosphere is created only at the divergent plate boundaries known as seafloor spreading centers (or ridges for short). Most of the planet's oceanic lithosphere forms along the vast, continuous ~40,000 km-long network of mid-ocean ridges that divide the globe like so many seams on a sloppily-made soccer ball. The Mid-Atlantic Ridge (above) is a world-class example. The pint-sized Juan de Fuca ridge (right and below) off the coast of Washington is a remnant of the East Pacific Rise, a Pacific spreading center that rivaled the Mid-Atlantic Ridge before its mid-section subducted beneath California.
Significant amounts of oceanic lithosphere also develop at smaller but often more complex spreading centers located in backarc basins near subduction zones and in maturing continental rifts like the Red Sea. Since mid-ocean ridges are the cleanest examples of spreading centers, we'll describe one next, but underneath their sedimentary covers, if any, all spreading centers look and work pretty much alike.
Think of a ridge as a snaking slit-like window exposing ready-to-melt asthenosphere to the hydrosphere. Mafic asthenospheric melts well up along the axial valley of the ridge to freeze against the edges of twin slabs forming on each side. In the map at right, the Pacific slab on the west is paired with much smaller twins across the Explorer, Juan de Fuca and Gorda ridges. South of the Mendocino fracture zone, the Farallon plate, a much larger Pacific plate twin, has already subducted beneath California along with the intervening ridge, the East Pacific Rise.
Submarine fissure volcanoes venting pillow basalts dot the axial valley of the spreading center, but down deep in the ridge, gabbro intrusions take over. Together, the basalt and gabbro make up the igneous basement of the oceanic crust. Ultramafic lithospheric mantle forms simultaneously beneath the basement as chilled asthenosphere freezes to its underside in a process known as underplating. Underplating progressively thickens the slab's mantle layer, but the basement layer changes little during the slab's time on the surface.
As they form, the twin slabs record the earth's current magnetic field as they move away from the ridge and from each other in more of less symmetrical fashion. They develop tell-tale magnetic stripes in the process. What drives the divergent plate motion across the ridge is still not entirely clear, but gravity appears to be the prime mover via a combination of hinge rollback and ridge slide. Superheated brines circulating through the fault-ridden ridge chemically alter the fresh slabs as they pass off the ridge and onto the adjacent ocean floor. Basted basalt, anyone?
Once formed, the twin slabs go their separate ways, eventually to meet similar if not simultaneous fates in the most fundamental cycle in plate tectonics. As they cool, they'll soon become denser than the asthenosphere on which they rest, but they'll manage to stay topside for a while, mostly via lateral compression by adjacent lithosphere. Their residence times at the surface will be brief by continental standards—only ~100 Ma on average. Ultimately, they'll sink into the upper mantle from whence they came at separate and oppositely polarized subduction zones that may or may not have existed at the time of their birth. After piling up and dissolving back into the upper mantle along the "660", their materials will reappear at the surface as new oceanic lithosphere generated at some other ridge, usually one or more continents away from the ridge of origin.
The word "center" notwithstanding, spreading centers are linear, not radial, seafloor features. They are segmented and offset at varying intervals by perpendicular linear seafloor features known as fracture zones or transforms (which are not quite the same as transform boundaries). Ultimately, fracture zones serve both to fit linear spreading center segments to the earth's spherical surface, and to allow them to tap the ripest asthenospheric magma sources available along strike.
Although they do show some seismicity, fracture zones are neither fractures nor continuously moving faults. Rather, they are welded sutures between slab segments generated at adjacent offset ridge segments. Because fracture zones may offset ridge segments by hundreds of kilometers, the slabs across them may differ substantially in age and therefore also in depth. The Mendocino fracture zone (right) separates seafloor segments differing in age by ~15 Ma, with scarps over 1 km high across the fracture zone. In the map at right, the Mendocino fracture zone offsets the Gorda ridge (off the coast of Oregon) from an adjacent segment of the East Pacific Rise already subducted beneath California south of the fracture zone.
Basaltic ocean islands and seamounts, in singles and in chains, and submarine plateaus dot the ocean floor all over the globe. They form atop standard-issue oceanic crust at particularly leaky spots along seafloor spreading centers, at oceanic ridge-ridge-ridge triple junctions, and also along extensional failures in the oceanic lithosphere.
Some plateaus approach subcontinental dimensions. Most plateaus remain submerged, but Iceland (right) is an intensely active emerging basaltic plateau straddling an anomalous segment of the Mid-Atlantic Ridge.
Topographically, so-called aseismic ridges look much like mid-ocean ridge segments, but they show no evidence of active ocean-floor spreading; hence the name "aseismic". Two such ridges emanate from the Iceland plateau, one connecting it to Greenland and the other underpinning the Faeroes. Aseismic ridges are also likely to be extensional features. A particularly large aseismic ridge is being consumed in the Nazca-South American subduction zone off Peru; its buoyancy is likely the cause of the Nazca's shallow subduction angle beneath Peru.
The 6,000 km long Hawaii-Emperor seamount chain (below) is probably built on a propagating extensional flaw in the Pacific plate.
Oceanic islands, volcanic seamounts and basaltic submarine plateaus mark many of the planet's hot spots—focused areas of excessive volcanism like Iceland, Hawaii (right), Afar, the Galapagos, the Azores, Yellowstone, the Snake River Plain, etc. Counts vary, but ~110 hot spots have been recognized in the last 10 Ma, many of them still active. Most hot spots develop on or near plate boundaries, but Hawaii and Yellowstone are notable midplate exceptions. One of the most prolific hot spots ever known and certainly the most active today, the ^Hawaiian hot spot (angled trace on the left in the map at right) has been punching up seamounts in the center of the Pacific plate for at least 70 Ma.
Hot spots probably have shallow roots in the upper mantle, but many are still commonly viewed as surface-penetrating melts induced by narrow plumes of heat originating deep within the lower mantle. Plumes are said to move slowly if at all relative to the bulk earth and are often invoked as a frame of reference for plate motions.
Originally proposed in 1971 to explain the Hawaii-Emperor seamount chain and the steady southward age progression of its lavas, the mantle plume concept was soon applied to many other hot spots. Plumes have recently run afoul of some well-established geophysical observations, however. For one thing, lower mantle processes can't generate direct surface expressions with an impenetrable barrier to material transfer in place at the upper-lower mantle boundary.
Then there's that pesky 60° angle between the south-trending Emperor chain and the southeast-trending Hawaiian chain (right), both of which owe their existence to the same long-running ^ Hawaiian hot spot. To explain the angle, plume models require the Pacific plate, the planet's largest, to make a sudden 60° turn relative to the "fixed" Hawaiian plume around 45±2 Ma, the age of the volcanic rocks at the bend. That's a lot of mass to redirect, but the Indian subcontinent first collided with the Eurasian plate about then, and a rapid global reorganization of plate motions might well have ensued. In many well-studied parts of the Pacific basin, ridge-generated patterns of seafloor magnetism should have recorded such a direction change, but they show the Pacific plate holding a steady course instead. This observation alone disproves the plume explanation for the Hawaii-Emperor chain; more technical arguments disprove plumes elsewhere.
Few of the many other volcanic chains adorning the Pacific floor show anything resembling a steady age progression along strike, so the Hawaii-Emperor is something of a fluke in that regard.
With the help of radiogenic heating and the insulating effect of continental lithosphere, heat escaping the lower mantle warms the upper mantle, while subducted slabs, cratonic keels, and mid-ocean ridges cool it. Since self-organizing gravity-driven plate motions control the positions of the continents, ridges and subduction zones, they also control the lateral temperature variations in the upper mantle. These tectonic temperature fluctuations are more than sufficient to account for all of the hot spots actually involving elevated surface heat flows. (Most are no hotter overall than their surroundings.)
Several lines of geophysical evidence point to normal lateral temperature variations of ±200°C in the asthenosphere and adjacent upper mantle around a global mean of ~1,400°C. None of the hot spots now recognized shows or requires a greater upper mantle temperature spike than that, and most show asthenospheric temperatures well under 100°C above the mean.
Most hot spots appear to be located by inter- and intraplate extensional stresses and driven by the eagerness of the asthenosphere to break out wherever and whenever it can. If the underlying asthenosphere is a bit hotter than usual, so much the better, but thermal anomalies are the exception rather than the rule and are by no means required.
Continental lithosphere is stiffer and thicker than oceanic lithosphere—often 100-200 km thick and up to 250 km deep along cratonic keels. Most of the planet's stock of old stable continental lithosphere was created during the Archean Eon by earth processes no longer active. Since then, new continental lithosphere has been created primarily if not entirely along convergent plate boundaries during subduction zone collisions, which have a tendency to convert to trapped oceanic lithosphere and its arc-generated and ridge-generated crustal embellishments into continental lithosphere. (We owe much of Colorado's 1.7 Ga basement to that process.)
Continental lithosphere can't be consumed in large amounts anywhere because it's far too buoyant to flush down a subduction zone (try pushing a basketball underwater), but small quantities can be stuffed under the edge of another continental plate—as at the Himalayas along the convergent Indian-Eurasian plate boundary and at the ^Alps (right) along the convergent Adriatic-Eurasian plate boundary. Continental lithosphere caught up in convergent plate boundaries tends to compress, thicken, heat, metamorphose, fault and rise.
Once formed, buoyant continental lithosphere strongly resists recycling into the upper mantle, but continents as planetary surface features are by no means permanent. Like political coalitions, continents are merely transient assemblages of continental lithosphere, constantly rafting together, breaking apart and regrouping—sometimes along similar lines, sometimes along vastly different lines. All ocean basins ultimately result from successful continental rifting. It is often the first step in the formation of sea floor spreading centers, but rifts sometimes fail (die out) before an ocean can form. Just as all slabs eventually subduct, all continents eventually rift.
Let's take a look now at two successful continental rifts floored with oceanic crust. In the maps above and at right, are the north northwest-trending Red Sea and the northeast-trending Gulf of Aden. They meet the developing south-trending East African Rift at a ridge-ridge-rift triple junction and hot spot in the Afar region of Ethiopia and Somalia. Whether the rift will eventually advance to the Red Sea stage, only time will tell.
Note that the Afar region is a hot spot only in the strict sense that excessive volcanism and seismic activity concentrate there. There is no evidence of unusually hot mantle beneath Afar. Extensional failure of the African plate is the root cause of all the commotion here, not a primary mantle distance.
A less dramatic but much more common form of rifting occurs in continental backarc settings, where it goes by the name backarc spreading once ocean floor appears. At most subduction zones, hinge rollback requires overriding plates to extend to maintain contact with the falling slab. Backarc rifting is the most common response. Depending on how far they progress, backarc basins can accumulate terrigenous rift, volcaniclastic and shelf-like marine sediments. When they progress to seaways like the Sea of Japan, the arc can migrate away from the mainland. When subduction zone collisions occur, as they often do over the life of an arc, backarc basins often invert into complex mountain masses composed of complex collisional crust as the outboard arc is shoved back against the mainland. Repetitive accordion-like backarc rifting and closing has long been important mechanism for the generation of continental crust in mobile belts.
Why continents rift apart is a question under active geophysical and geologic investigation, but regardless of the reason, large continents appear to be inherently unstable over timescales of 100 Ma or so. Since all plates are weakest in extension, this comes as no great surprise, but the nuts and bolts of the process remain unclear.
An oft-proposed mechanism for continental rifting with few compelling examples is the build-up of lithosphere-weakening heat under large, thick continents acting as thermal blankets. Since upper mantle beneath the interior of a large continent can't cool itself by subduction or sea-floor spreading, as it normally does, it could, if denied adequate circulation, become hot enough over time to soften, elevate and eventually rift the overlying plate. Lacking the refractory keels that insulate the cratons, mobile belts (like Colorado?) would be particularly susceptible, but the asthenosphere may convect fast enough to prevent most if not all continent-splitting heat build-ups. A growing body of evidence indicates that cold-start rifting related to certain concentrated crustal loads occurs more often than hot failures of continental lithosphere.
With the help of gravity, continents adjacent to mixed plate boundaries that are partly divergent can also literally fall apart. The western third of the Unites States has been doing just that for ~30 Ma as the Pacific plate pulls away from North America there. This particular mobile belt break-up is both distributed over the entire Basin and Range and concentrated in the Rio Grande Rift of New Mexico, Colorado and Wyoming.
Whatever the cause, the continents have been rifting and regrouping in a supercontinent cycle since the onset of plate tectonics around ~2.0 Ga, with no end in sight.
Most plates include both oceanic and continental lithosphere meeting along a solid and relatively quiet suture known as a passive continental margin. Such margins always represent one side of a continental rift that succeeded in opening an ocean basin. The Atlantic coasts of North America, South America, Europe and Africa are major passive margins along the sides of the same successful rift, now the Atlantic Ocean and still spreading along the Mid-Atlantic Ridge (right). Over the lifetime of an ocean, passive margins like these can accumulate very thick sedimentary piles eventually developing into broad continental shelves resting for the most part on continental lithosphere riddled with normal faults.
Basaltic volcanism and active normal faulting are the rule at developing passive margins. The volcanism eventually ends as passive margins mature, but the seismicity last much longer. As the growing continental shelves progressively load up the fossil rift faults that gave birth to the adjacent ocean basin, upper crustal adjustments in the form of passive margin earthquakes like the 1886 ~M7.3 Charleston, SC event result.
Continents have been clumping and splitting since the inception of plate tectonics at ~2.0 Ga. The process has operated at all scales to produce supercontinents (unions of two or more continents), microcontinents (slivers of continental lithosphere produced in a variety of ways), and everything in between. Cratons, the seemingly indestructible continental cores formed in Archean time, have remained intact throughout but have recombined freely with other cratons over time.
At the extreme, the planet has twice seen all of its continental lithosphere gathered into a single landmass:
There may have been earlier single-continent episodes as well.
Lesser multi-continental rafts have formed with greater frequency. Gondwana tied up today's southern continents from ~650 Ma to ~130 Ma. In between, it joined and left Pangea. North America was trimmed out of Laurentia, a northern supercontinent that once included Baltica (northern Europe) and the eastern part of Gondwana. Today's largest continent, Eurasia, combines earlier continents sutured at the Ural Mountains. India punched in at ~ 45 Ma, Africa has been docking since ~80 Ma, and Australia is bearing down on it as we speak.
Why unsinkable continents collide and bond temporarily is not hard to understand on a planet whose surface dynamics are dominated by subduction. Why they rift apart again is less clear, but upper mantle heat build-up related to the insulating effect of continental lithosphere and isolation from the cooling effect of subduction may play a role. Why the continents keep gathering into large supercontinents is also a mystery, but the known frequency is higher than one would expect from random plate motions alone—hence the concept of a supercontinent cycle. Some theorize that supercontinents collect over large-scale mantle downwellings, but we have no evidence that such downwellings have ever occurred. Nor are the likely in an upper mantle constrained to circulate above the lower mantle at 660 km.
Lithosphere is usually either distinctly continental or decidedly oceanic, and mostly the latter, but mixed situations occur now and then, usually in a rift setting. For example, new oceanic lithosphere developing at the bottom of an advanced continental rift or backarc basin can easily accumulate several kilometers of silica-rich sediments shed from adjacent continental crust while the rift is still narrow. If the rift fails at that point, a sliver of mixed lithosphere is left embedded within an otherwise continental plate. If the rift goes on to form a world-class ocean basin, the continental shelves at its margins will accumulate even thicker piles of continent-derived sediments resting in part on oceanic lithosphere. If the rift is a backarc or ocean basin that later closes in a collision, the lithosphere formed there will be primarily continental in nature with a very complex crust.
Skip to Subduction Zone Collisions
Subduction occurs when one plate slips under the edge of another at a convergent plate boundary. The lithosphere on the down-going side is almost always oceanic because its negative buoyancy promotes subduction, but positively buoyant continental lithosphere occasionally tries to go under as well, often with spectacular results like the Alps or the Himalayas.
It has only recently been recognized that self-organized gravity-driven subduction, not mantle convection, has been the mighty engine powering plate tectonics since its inception at ~2.0 Ga. Furthermore, subduction largely drives the limited convection found in the upper mantle, not the other way around.
And how does subduction rate the leading role? Ultimately, because heat loss and gravity are the only important motive forces acting on the plates. Continuing top-down cooling of the upper mantle produces gravitationally unstable oceanic lithosphere that progressively cools, thickens, stiffens, imbibes water and becomes more dense through cooling and phase changes as it rafts away from its spreading center of origin. Not far from its spreading center, the oceanic lithosphere (slab for short) is already denser and stiffer than the hot asthenosphere on which it rests. Flexural support and lateral compression from surrounding slab segments may allow it to stay topside for a while despite the density inversion, but eventually, all overdense slabs buckle downward and sink subvertically into the upper mantle along a convergent plate boundaries known as a subduction zones. Metamorphic reactions occurring ~60 km below the surface make the crust of the falling slab even denser, thereby speeding its fall.
To understand how lateral compression can prevent an overdense slab from sinking right away, think of some wooden blocks lined up on a table. If the line isn't too long, you can squeeze its ends together and lift it off the table as a unit, but the longer the line, the harder you'll have to squeeze. Add a little weight to the center of the longest line you can lift intact, and it will promptly collapse—i.e., the center will fall under gravity. To model the contribution of flexural support, you could add glue between the blocks, but with a long enough line, or enough added weight, that would only delay the inevitable. If the glued joints proved stronger than the wood, the blocks themselves would break to allow the collapse. Gravity will not be denied.
Slabs behave in much the same way. Once they become dense enough with age, they must fall. When they do, a subduction zone appears.
The cross-section through the Java trench and the Sumatra arc at right is well-constrained by geophysical data and is drawn without vertical exaggeration. The following features are typical of real subduction zones but not of popular subduction diagrams (which are often drawn inaccurately with very misleading vertical exaggerations):
Now let's work our way from the slab and trench toward the magmatic arc.
On average, oceanic plates are 80-90 km thick when they finally begin to sink in earnest at a subduction zone. The actively sinking portion of a subducting slab bends down into the asthenosphere well beneath, not in front of, the edge of the overriding plate. The bend between the sinking and about-to-sink slab segments is called the hinge. Beyond the hinge, most slabs dip into the mantle at 30-70° angles, but important exceptions (like the low-angle Nazca plate beneath the western margin of South America) occur.
Once the slab reaches a depth of ~60 km as it sinks past the hinge, the basalt and gabbro making up its crust undergo phase changes to much denser eclogite, a high-pressure metamorphic rock 200-250 kg/m3 denser than upper mantle rock. The eclogite acts like a lead sinker, speeding the fall of the slab. Water released into the mantle wedge overlying the slab as a chemical byproduct of the eclogite transition eventually triggers the partial melting that fuels arc magmatism.
It's important to keep in mind that slabs don't really punch into the mantle, as diagrams and popular animations often depict. Instead, the sinking segment beyond the hinge falls broadside, like a paddle, toward the 660—the global upper-lower mantle boundary at 660 km. Slabs do slide horizontally through the hinge region to some extent, but their dominant motion at and beyond the hinge is not a slide but a near-vertical drop.
Falling slabs often break up and may even fold up on the way down. Unable to sink beyond the impenetrable 660, sunken slab fragments pile up there junkyard-like to await recycling into the upper mantle, eventually to reappear as melt at the spreading centers where new oceanic lithosphere forms.
This fundamentally oceanic cycle has operated for the last ~2.0 Ga with no end in sight. Indirectly, it also acts to redistribute the planet's fixed fund of continental crust and its slowly growing fund of continental lithosphere.
Over time, most trenches accumulate accretionary wedges—fault-riddled jumbles of oceanic crust scraped off the falling slab mixed with sediments shed off the overriding plate. (Think of the mass of dirt rolling before a bulldozer blade.) Wedges usually rest on oceanic crust in front of the leading edge of the overriding plate. Mature wedges keep the angle between the walls of the trench fairly flat—usually in excess of 150°.
In most subduction zones, accretionary wedges effectively caulk, pad and lubricate the physical contact between slab and overriding plate. In the process, they generate most of the world's earthquakes, including nearly all of the ^very largest. (The two largest shocks ever recorded, the ^May 27, 1964 M9.2 Alaskan earthquake in Prince William Sound and the ^May 22, 1960 M9.5 Chilean earthquake, are prime examples.) Only in the sediment-starved trenches of small open-ocean arcs do the falling slab and the overriding plate actually touch. Padding by the wedges probably contributes to the rarity of forearc damage at subduction zones.
Between the trench and the arc, in the arc-trench gap, lies the forearc region. Basins filled with volcaniclastic sediments often form atop the forearc; below and trenchward is the accretionary wedge.
Interestingly, forearcs rarely if ever show evidence of damage due to either compression or tectonic erosion along their contacts with falling slabs—apparently because most trenches are kept in tension by hinge rollback. Even subduction zones with fast convergence rates exceeding the rollback rate rarely show forearc damage. The Nasca-South American boundary is such a subduction zone, and its forearc basins remain undeformed even as the shallow-angle slab couples with and deforms the Andes and the eastern foothills at and behind the Andean arc.
Another important wedge at the subduction zone is the mantle wedge situated between the top of the falling slab and the bottom of the overriding plate. Low-viscosity asthenosphere wells up into the mantle wedge as the slab falls away from the overriding plate. The mantle wedge provides the melts that surface in the magmatic arc. In continental arc settings, it also heats the backarc region via conduction. If the backarc is actively rifting and thinning, the underlying wedge can undergo partial decompression melting. These typically basaltic melts further heat the backarc as they rise through it.
Gravity-driven hinge rollback keeps most convergent boundaries under tension, not compression; yet, gaps between plates are never observed. At every subduction zone involving a falling slab, the edge of the overriding plate somehow manages to hover unwaveringly over the bend in the falling slab. Most overriding plates, oceanic and continental alike, extend as needed to maintain contact with the free-falling slab, usually via backarc or intra-arc spreading.
The forces driving this extension remain controversial, but they must be irresistible, because they never seem to fail to maintain contact. Some authors propose a viscous coupling that goes something like this: The falling slab draws overlying asthenosphere toward it, into the mantle wedge. The entrained asthenosphere in turn drags the edge of the overriding plate over the slab hinge. Since it's easier to move the edge than the entire overriding plate, and since plates are weakest in extension, the edge pulls away until the forearc comes to rest gently on the slab. Gravity acting on the overriding plate probably figures in somewhere.
In the map above, backarc spreading in the Sea of Japan and Sea of Okhotsk allows the Japanese and Kuril Islands to extend toward and remain in contact with the subducting Pacific slab.
Subduction-related compression of overriding plates is the exception rather than the norm. Prior to 29 Ma, California, a typical case in this regard, hosted a subduction zone continuously active for over 100 Ma with no significant forearc shortening. If compression had ever developed across this long-lived subduction zone at the point of initial plate-to-plate contact, the forearc would have been the first structure to deform.
Overriding plates crumple due to subduction only under one or more of the following uncommon circumstances:
In many such cases, the subducting slab ends up falling at a shallow enough angle to couple mechanically with the overriding plate well beyond the trench. Often, overriding plate compression involves a combination of these factors. The North American Laramide Orogeny probably involved a combination of the last two.
Active subduction zones involving a falling slab also include a magmatic arc, a band of intense volcanic and intrusive activity affecting the overriding plate. When the top of the falling slab reaches a depth of 80-120 km, partial melting begins to occur in the barely solid hot mantle wedge just above the slab. These first melts rise to the surface along the arc's volcanic front. In general, the greater the dip of the falling slab, the sooner the slab reaches the first-melt depth contour, and the narrower the arc-trench gap. Melts and mantle heat also rise along a band of varying with behind the volcanic front known as the backarc. Extension and rifting of the backarc work to accommodate hinge rollback.
Exactly how arc melts develop in the mantle wedge remains controversial, but everyone agrees now that they don't involve melting of the subducting slab itself. One promising recent proposal involves serpentinite as a important intermediary. As slabs fall, they release large amounts of water from entrained sediments and also as a byproduct of the metamorphosis of their basalt and gabbro to eclogite at ~60 km. The rising water reacts with peridotite in the immediately adjacent mantle wedge to form a layer of serpentinite that rides along the top of the slab. As long as it remains stable, the serpentinite effectively shields the mantle wedge above by absorbing the watery fluids rising from the slab. But when it reaches a temperature of 600-650°C, usually at a depth of 80-120 km, the serpentinite decomposes and releases its water into the overlying mantle wedge. At these depths, these secondary watery fluids induce partial melting of the adjacent mantle wedge by reacting with anhydrous mantle minerals to form hydrated minerals with melting points below the ambient mantle temperature. The resulting partial melts rise toward the surface through the mantle wedge and the overriding plate to form the magmatic arc.
Bear in mind that the arc is not a fixed feature of the overriding plate. Rather, its location tracks the depth of the subducting slab beneath it like a gigantic contour line drawn with volcanoes. Since the slab begins to generate ^magma (melt) only when its top lies 100±20 km below the surface (i.e., when its serpentinite crust decomposes at 600-650°C), the location of the volcanic front relative to the trench changes as the dip of the falling slab varies. As the dip of the Farallon plate subducting along the West Coast shallowed out progressively during the Laramide Orogeny, associated arc magmatism migrated rapidly to the east from the Sierra Nevada, its original locus, to the Rockies. When the entire slab rose above the melt-generating depth, arc magmatism snuffed out.
Since deeper slab segments also induce asthenospheric melts, arc magmatism affects a band of variable width behind the volcanic front. In general, the greater the dip of the falling slab, the sooner the slab reaches the first-melt contour, the narrower the arc-trench gap and the narrower the magmatic arc behind the front.
Active volcanism is a common feature of magmatic arcs, as anyone from Oregon, Mexico, Chile, Japan, Fiji, Indonesia or the Philippines will tell you, but subsurface intrusions are also important in arc evolution. At right above is Mount St. Helens, part of the Southern Cascades arc. Below it is a magma chamber that may well freeze in place one day to become an intrusive component of the arc.
At any given point along an evolving subduction zone, the overriding plate may be continental, oceanic or something in between, as at ^Augustine Volcano (right), an active vent in the eastern Aleutian Archipelago fed by melts generated by subduction of the Pacific plate beneath the Alaskan portion of North America. Each overriding margin type responds to subduction in its own way, but in every case, a magmatic arc eventually forms a variable distance back from the trench.
Arc magmas tend to be basaltic at young oceanic arcs, andesitic at mature arcs and rhyolitic or dacitic at continental arcs. The initially basalt-like mafic melts feeding magmatic arcs come not from the sinking slab, as popular diagrams commonly show, but from a complex hydration process in the mantle wedge above it (the red zone at far bottom right in the diagram at right). As the proto-magmas rise from their mantle source region, they interact with surrounding mantle and crustal materials in a number of ways, including:
By the time they reach the surface, some asthenospheric melts are still quite basaltic, particularly at small, primitive open-ocean island arcs, but they may also become quite felsic in more mature oceanic and continental arcs. Arcs that start out spewing primitive runny mafic basalt and end with explosive eruptions of sticky, highly felsic rhyolite are said to be bimodal, and they're fairly common. However, the average arc lava, andesite, falls somewhere in between, both in composition and viscosity. Dacite is another common felsic product intermediate in silica content between andesite and rhyolite. Viscous andesitic, dacitic and rhyolitic magmas charged with ^gas tend explode on reaching their vents, especially the high-silica dacites and rhyolites. Low-silica basaltic magmas tend to erupt much more peacefully, as on the island of Hawaii.
At right is ^Mt. Pinatubo, a notorious Philippines island arc stratovolcano with predominantly andesitic and dacitic output located on the island of Luzon. Glowing cloud eruptions and volcanic mudflows (lahars) ravaged Luzon and killed thousands during Pinatubo's vicious June, 1991 eruption sequence. The June 12, 1991 explosion shown here sent up a towering vertical (Plinian) ash column and injected heavy loads of sulfur dioxide into the stratosphere. In the culminating cataclysmic eruption of June 15, 1991, Pinatubo lost 500' of its 5,275' pre-eruption height.
Subduction-related magmatic arcs built on overriding continental plates tend to erupt explosive rhyodacitic magmas. Modern examples include the the Andes of Chile and Peru (right), the Southern Cascades of Washington and Oregon (shown in cross-section in the diagram above) and the Apennines of Italy.
The Andes arc remains fully active as the Nazca plate flooring the SE Pacific basin subducts rapidly beneath the west coast of South America. The active Cascades arc draws fire from the small Juan de Fuca plate still subducting along the Washington and Oregon coasts. The Sierra Nevada mountains of California are the uplifted and eroded fossil roots of a now abandoned continental arc active off and on at ~145-30 Ma, when the Farallon plate of the Pacific basin subducted under the west coast of California.
As they evolve, continental arcs absorb and erupt subduction-induced magmas of increasingly felsic compositions (high in silicon and aluminum and low in iron and magnesium) as the rising magmas entrain more and more sialic (silicon- and aluminum-rich) material during their passage through lower continental crustal levels. The higher silica concentrations lead to more explosive eruptions, as at Mt. St. Helens on May 18, 1980 (right). Accretionary wedges along continental arcs are typically large due to generous sediment influxes from a large and typically mountainous eroding coastal landmass.
Continental margins at subduction zones extend toward the trench to keep up with hinge rollback. This extension is usually localized to the backarc, the heat-softened magmatic belt just behind (inland of) the volcanic front. The backarc is doubly heated—from below by mantle wedge circulation and from within by magmatism that strays behind the volcanic front. As extension progresses, rifting will appear within either the arc or the backarc to form a backarc basin. As they develop, continental backarc basins accumulate varying volumes of terrigenous rift, volcaniclastic and even shelf-like sediments, depending on the setting. If a subduction zone collision doesn't close the basin first, the basin will eventually become a seaway that pushes the arc front away from the mainland via seafloor spreading, as the Ryuku, Japanese and Kuril arcs have done at right.
Subduction-related oceanic or island arcs built on overriding oceanic plates tend to erupt basaltic to andesitic magmas, depending on their maturity. Modern examples include the Tonga-Kermandec Islands east of Australia, the Philippines, and the ^Lesser Sunda Islands (right, Timor is the largest) of Eastern Indonesia north of Australia.
Island arc volcanic and intrusive igneous rocks tend to be more mafic than those of continental arcs because their magmas pass though more mafic lower crust on their way up. Island arcs are more flexible than continental arcs and can more easily reshape themselves as conditions along their subduction zones change. Accretionary wedges along island arcs are usually smaller than their continental counterparts.
Open-ocean island arcs also undergo backarc spreading to accommodate hinge rollback, but they accumulate sediments . Many of the curious double arcs of the western Pacific basin have been split by backarc spreading.
Any oceanic arc over 64 km (40 mi) wide is likely to be a composite of several arcs sutured together in the wake of one or more arc-arc collisions. In the process, composite arcs tend to develop significant amounts of continental crust. Notable examples include Sumatra (right), New Zealand's south island and the Philippines and the Japanese islands.
Slab fall is by far the dominant drive for plate tectonics, but it can play out on the surface in surprisingly complex ways, particularly due to a phenomenon known as hinge rollback. The diagram at right would be more accurate if it showed the slab beyond the hinge falling faster than it dips into the mantle and the hinge steadily moving back toward the trench.
As the dipping edge of a foundering slab falls toward the "660", its hinge rolls back (retreats) into the segment still topside and draws the overriding plate over its hinge in the process. The overriding plate stretches or moves toward the retreating hinge, as its situation and material properties allow, in order to maintain contact with the subducting slab. (Gravity, lower crust ductility in the overriding plate and asthenospheric mobility would all work to fill any gap that might otherwise develop at the trench.)
This well-documented hinge-chasing behavior of overriding plates is a critical element in the latest formulation of plate tectonics. It plays a crucial role in driving plate motions and in the development of ocean basins and continents alike.
A second minor way in which a slab might fall subhorizontally is called ridge slide. As a slab moves away from its ridge and cools from the top, the lithospheric mantle along its underside thickens by ~100 km due to underplating by cooling asthenosphere. The viscous boundary between the slab and the underlying asthenosphere thus forms a shallow ramp sloping down from ridge to subduction zone. Slabs might theoretically slide down this ramp under gravity, but direct evidence of ridge slide is lacking.
Contrary to popular accounts, there is little if any "ridge push" at spreading centers (the nascent oceanic lithosphere there is too thin, hot and floppy to push anything around), and little if any "slab pull", since the foundering slabs sink subvertically rather than slide sideways into the upper mantle and probably lack the required tensile strength to drag along distant topside segments anyway. Nor is there credible evidence that upper mantle convection currents push cratons around by their keels like so many upside-down sailboats.
Much of the malleability displayed by subduction zones over time is accomplished by backarc spreading modulated by along-arc variations in hinge rollback. If the arrival of a buoyant ocean floor edifice temporarily slows or halts hinge rollback and overriding plate extension at one point along a long subduction zone, hinge rollback will "festoon" (Warren Hamiliton's nicely visual word) around the tack point, while the overriding plate extends around the compressive section to follow the altered hinge geometry.
Iwo Jima and the Volcano Islands stand above a notch (tack point) in the long, continuous Izu-Mariana trench east of Japan (above). The largest remaining seamount in an east-west chain now subducts at the notch. South of the notch, backarc spreading widely splits the Mariana arc (right) to keep pace with unfettered hinge rollback away from the notch. The Bonin arc north of the notch has split as well. Both splits pinch out at the notch (right), which first and foremost marks the local absence of rollback and extension.
Rollback engenders similar splits along continental arcs. The Yellow Sea, the Sea of Japan and the Sea of Okhotsk all occupy well-developed backarc basins developed along the west coast of mainland Asia behind the Ryuku, Japanese and Kuril arcs, respectively. The Yellow Sea's large continental shelf collects mature sediments from several major rivers draining much of China. Such sediments would not be found in the backarc basin of an island arc.
A more diffuse style of continental backarc extension caused the initial (but not current) stage of normal faulting in the Basin and Range of the western United States between 40 and 29 Ma—after post-Laramide resumption of normal-angle subduction and hinge rollback along the West Coast, but before subduction of the East Pacific Rise and the cutting of the San Andreas fault.
Finally, note the prominent notch in the west coast of South America and the highest segment of the Andes surrounding it. One theory for the notch posits that asthenospheric drag on the continental keel pulls the entire midsection of the South America to the east relative to the rest of the continent, notch included. I wonder about the two aseismic ridges entering the trench at the the notch. Much of the Nazca-South American subduction zone around the notch is currently in compression due shallow slab subduction, which is in turn due to a combination of faster than usual plate convergence and the subduction of buoyant ocean floor edifices like the Nazca Ridge off Peru. Could extension of other segments of the west coast have played a role in shaping the notch, perhaps in the past? I hope to research this question.
Skip to Intraplate Deformations
Even a casual glance at a shaded relief map of any ocean basin sans water will convince you that the oceanic crust is far from smooth, even over the abyssal plains that make up the bulk of the world's ocean floor. In fact, the ocean basins have more relief on average than the continents. As the map of the Costa Rican segment of the Middle America Trench at right shows, the ocean floor bristles with
The larger bumps don't always go down easily at subduction zones, particularly if they contain a lot of buoyant arc-generated or continental crust. When a sufficiently large or buoyant incoming continental fragment or oceanic edifice (like an arc or a submarine plateau) jams into a subduction zone, compressive stresses wrack the overriding plate, often with sufficient force to close and even invert any backarc or intra-arc basins that may have formed due to earlier hinge rollback.
Open-ocean island arcs often wrap themselves around such obstructions via variations in hinge rollback, but continental arcs are more inclined to shorten along collision zones. Through tectonic burial, temperature- and pressure-related recrystallization and compressive uplift, the jamming edifice, surrounding trench fill, nearby arc materials and any associated inverted backarc basins often end up as elevated metamorphic crust after the collision, no matter how deep they may have been dragged in the throes of the collision.
Buoyant but less obstructive slab features may instead flatten the angle of subduction beneath the overriding plate. In such cases, the flattened slab often couples mechanically with the overriding plate well inboard of the trench.
The map at right shows typical seafloor topographic variation along the southern Middle America Trench, where the Cocos and Nazca plates subduct beneath the Costa Rican magmatic arc built on the oceanic southwest margin of the Caribbean plate. At the top of the map, the subducting Cocos slab off Nicaragua is fairly smooth down to Costa Rica's Nicoya Peninsula (the largest of the horsehead-shaped salients). Further south, the Cocos plate bears many seamounts and submarine plateaus. Their buoyancy flattens the dip of the subducting slab beneath central Costa Rica, where the magmatic arc bows northeast away from the trench as it follows the 100 km depth contour of the falling slab. Further south, the thick crust of the Cocos Ridge enters the trench. The subducting seamounts also appear to control the location and style of Costa Rica's many large subduction zone earthquakes.
Arcs are less buoyant than continents but are still too light to subduct without a fight. Triassic through Jurassic arc-continent collisions along the west coast of the US deformed the overriding North American plate several times—among them the Antler, Sonoma and Nevadan Orogenies. These events were not felt not as far east as Colorado, but they added a good bit of land and rumple through at least California and Nevada. Some of the added terranes were later carted away to the north along major transform faults to take up residence in southwest Alaska.
When two continents collide along a subduction zone, the underthrusting continent goes down only so far, and world-class mountain ranges come up. Colliding continental sections of the Indian and Eurasian plates have been pushing up the Himalayas and the Tibetan Plateau (right) over the subducting Indian plate since 45 Ma. Most arc-related collisions are mere fender-benders by comparison.
The ongoing collision of the Eurasian and African plates began ~80 Ma ago. On the west, it's managed to fold up the Atlas and ^Anti-Atlas Mountains of Morocco (left) on the down-going African side. Further east, it's thrown up the Alps (right) on the overriding Eurasian side of the collision zone.
Oceanic edifices trapped in the closing ocean or sea between colliding continents often end up in the mix of elevated crust at the suture zone. The intervening swarm of oceanic arcs and small continental fragments in and around the ^Banda Sea of Eastern Indonesia (right) will make for some very complex collision-generated continental crust when northbound Australia and Southeast Asia finally meet.
J. Tuzo Wilson, one of the fathers of modern plate tectonics, recognized early on that ocean basins often open and close repeatedly along similar but not identical passive continental margins, presumably along bands of enduring lithospheric weakness. An oft-cited example of the Wilson cycle is the rifting of the supercontinent Pangea to form the current Atlantic basin subparallel to Appalachian and pre-Appalachian orogenic belts left by earlier closings of the Iapetus and Rheic Oceans in latest Proterozoic through late Paleozoic times. This accordion-style "split and come back together" continental regrouping is termed introversion. On the regrouping side, introversion makes sense to the extent that supercontinents collect over large-scale mantle downwellings that persist throughout their breakup, but then the breakups themselves become hard to explain.
Recently, the Wilson cycle has been expanded to a "supercontinent cycle" that also includes cases where supercontinents split and reorganize by closing external oceans rather than by reclosing internal oceans. This "split and stick to the walls" pattern is termed extroversion. The breakup of the supercontinent Rodinia ca. 1.1-0.6 Ga and the subsequent formation of Gondwana ca. 0.6-0.55 Ga appear to have involved just such an extroversion through the closing of portions of a large peri-Rodina ocean. One theory of extroversion has continental fragments sliding off a thermal (and physical) high developing in the mantle beneath the dispersing supercontinent. The fragments reconvene over a large-scale mantle downwellings marked by subduction zones. However, this explanation becomes difficult if subduction is purely gravity-driven and upper and lower mantle motions are decoupled, as this article favors.
As usual in earth science, the introversion and extroversion models are simply end members in a spectrum of possible supercontinent behaviors. Like it or not, mixed cases are probably the rule rather than the exception. For example, the formation of the Appalachian orogen turns out to be more complex than the simple Wilson cycle would suggest, as it now appears that the passive margin that separated from Laurentia (the main North American precursor) to open the Iapetus ocean was not the margin that later collided with Laurentia when Iapetus closed.
For the continents to clump and split, some oceans must grow while others shrink. The modern Atlantic is a growing ocean with one very small subduction zone at the Lesser Antilles. Elsewhere, its large twin slabs end against passive margins and transforms with no other consumption to offset spreading at the Mid-Atlantic and peri-Antarctic ridges. Larger Atlantic subduction zones will develop when its slabs are ready to fall in earnest, but for now, the slabs are managing to stay topside.
The Atlantic grows largely at the expense of the Pacific, a much larger but shrinking ocean that continues to lose area to hinge rollback at its many major subduction zones even though its ridges post some of the fastest spreading rates on earth. As North America passes over upper mantle once beneath the Pacific, material recycled from long-gone Pacific slabs reappears in new Atlantic slab segments.
If the Pacific eventually closes this time around, we'll be able to say that Pangea ended in extroversion like Rodinia before it. If the Atlantic closes instead, we'll have a 3rd introversion
A less obvious consequence of the supercontinent cycle is a global traffic in continental crust. Whether though the closure of internal or external oceans, the continued formation and subsequent breakup of supercontinents allows continents to exchange parcels of crust at their margins. The result is a slow, sliver by sliver redistribution of continental crust over the globe. Take the margins of the Atlantic basin (above right), currently is in its 3rd edition. Along the way, North America acquired pieces of Africa and South America, while Scotland and Norway came away with pieces of North America. For a stage-by-stage walk-through of the process, drop by Lynn Fichter's ^Wilson cycle tutorial.
Most of the globe's endowment of true high-silica continental crust seems to have formed in the Archean Eon through obscure earth processes never repeated since, but the continents continue to grow along their margins via subduction, which effectively rafts in or welds on more primitive (mafic) crust developed in the volcanic arcs and backarc basins surrounding subduction zones and in the large submarine basaltic plateaus and aseismic ridges erupted over oceanic ridge-ridge-ridge triple junctions and other leaky ridge segments. As subduction plays out in the ocean basins, continents are endlessly reorganized through rifting and collisions, through large-scale strike-slip faulting (as along the San Andreas and Queen Charlotte faults of the western North American margin), and through accretion of oceanic elements.
On ocean closure, ocean floor edifices too buoyant to subduct are swept up into the crust along the suture between approaching continents. When the continents next rift apart along roughly the same suture (as they sometimes do), this material, now metamorphic, will have been incorporated into one or both of the new passive continental margins. The supercontinent cycle ensures that at least some older continental materials also change hands as continents part and regroup through time.
Skip to References
Plate tectonics is a well-developed discipline only in the vicinity of plate margins and has only been operative, as best we can tell, for the last ~2.0 Ga. Geoscientists still struggle to understand the strong and repeated intracontinental deformations so beautifully wrought in Colorado and surrounding states (right).
As noted above, fully 15% of the earth's surface currently violates the basic plate tectonic model of rigid plates interacting and deforming only at their boundaries. Far-field plate tectonic stresses, gravitational collapse and mantle stirrings all appear to be involved in intraplate deformations.
Far-field stresses generated at distant plate margins can produce significant intraplate deformations, especially when they can exploit local weaknesses. The exact mechanisms remain under debate, but mobile belts, with their thinned lithosphere and many lithosphere-scale faults, are particularly susceptible to far-field deformation.
Modern Asia's topographic response to the ongoing Indian-Eurasion continent-continent collision is easily recognizable over a 2,000 km-wide swath roughly centered on the Tibetan Plateau (left). North of the plateau, hundreds of kilometers beyond the plate boundary near the Himalayan front, is a far-field fold and thrust belt itself hundreds of kilometers wide. And well beyond that to the northwest are western China's Tian Shan mountains, the product of complex a far-field strike-slip regime involved in Southeast Asia's eastward flight from the Indian-Eurasian collision path. To the east of the Tibetan Plateau, swarms of north-trending far-field strike-slip faults extending far into Southeast Asia accommodate India's penetration of the Eurasian plate.
A similar scenario may have developed in western North America around 300 Ma, when deeply intracontinental Ancestral Rocky Mountains (ARM) rose in Colorado in response to the Ouachita Orogeny, a continent-continent collision between North America and Africa that played out nearly 1,000 km to the southeast during the final assembly of the supercontinent Pangea. The particular basement-involved deformation style seen in the Ancestral Rocky Mountains lacks a good Himalayan analog, but the far-field tectonic settings are somewhat similar.
Colorado's next major far-field event, the Laramide Orogeny at ~72 to 40 Ma, is harder to explain. Most current investigators attribute the Laramide to east northeast-trending regional crustal shortening related to an episode of flat or very shallow-angle subduction beneath the western margin of North America, then some 800 km away. The Laramide ended with a devastating mid-Tertiary magmatic bang as the now departed Farallon Plate finally fell away from the underside of the continent to re-expose the lithosphere of the western US to an inflow of hot asthenosphere. A similar shallow-angle subduction scenario appears to be unfolding (or more accurately, folding) in the eastern Andes today in response to balky subduction of the Nazca Ridge (that large southwest-trending linear sea floor feature off southern Peru in the map at right) at the western margin of South America. The basement-involved style of deformation is a good match for the Laramide's, but this modern analog isn't as far afield. In any event, how the required forces are transmitted from the shallow slab to the shortened inboard crust remains a matter of much debate.
In order to produce major deformations the likes of the Ancestral Rocky Mountains and the Laramide Rockies at distances of 800 to 1,000 km, plate-generated far-field stresses usually have to reactivate (capitalize on) pre-existing crustal and lithospheric weaknesses. In Colorado, basement-penetrating Late Proterozoic rift faults and much older lithosphere-scale Early Proterozoic fossil sutures provide many such opportunities, and the rock record indeed confirms repeated activity on some, like the Gore fault along the west flank of the Gore Range. This Early Proterozoic discontinuity was clearly reactivated during both the ARM and the Laramide orogenies.
Since ~28 Ma, far-field extension and regional uplift have dominated the tectonic regime in Colorado — particularly along the Rio Grande rift. The Pacific plate's westward divergence relative to North America no doubt facilitates the extension in concert with gravitational collapse, but the large-scale domal uplift accompanying the extension is difficult to envision as a primary or even secondary far-field plate effect. Nor is it entirely explainable as the shoulder uplift that normally attends continental rifting. Mantle forces unrelated to plate tectonics may also be at work here.
A pile of sugar heaped beyond its sustainable angle of repose (~30°) slumps until its flanks are once again capable of self-support. Thanks to the lithosphere's very limited tensile strength, uplands lifted beyond their limits of internal support do much the same. Oversteepened regions like the Himalayan front (at bottom in the image above) and the Basin and Range of the western United States (left of center in the image at right) eventually fail and spread out under their own weight in a common process known as gravitational collapse. The Southern Rockies appear to be doing the same along the Rio Grande rift.
Normal faults rooted in low-angle detachment surfaces accommodate gravitational collapse at brittle upper crustal levels, while the underlying ductile lower crust slowly flows downhill. The Basin and Range province has extended in this manner by more than 100% in response to an over-steepening of Rockies and the Pacific plate's slow westward pull-away from North America along the San Andreas transform. East of the Colorado Plateau, the extension has topped 300%, in part to accommodate a ball bearing-like clockwise rotation of California's Transverse Ranges along the San Andreas fault.
Gravitational collapse may also operate at subduction zones, where overriding plate margins seem quite willing to stretch as needed via intra-arc or back-arc spreading to maintain contact with the falling oceanic slab.
As ever more sophisticated remote sensing technologies begin to unmask the stirrings of the earth below the lithosphere, it becomes increasingly likely that upper mantle processes ostensibly unrelated to plate interactions can also shape intraplate regions. In Colorado's case, the greatest lithospheric deformations follow the Rio Grande Rift and the Colorado Mineral Belt, the state's most profound structural defects. Upwelling of the upper mantle appears, at least in part, to drive the large-scale domal uplift and rifting centered at the intersection of these two lineaments near Leadville. Along the southwest flank of the dome, unusually numerous travertine springs dotting the floor of the Grand Canyon outgas anomalously high quantities of CO2 and 3He. The latter, a primordial isotope no longer produced anywhere in the solar system outside the center of the sun, can only come from the mantle. Why the upwelling occurs just here, and not elsewhere in North America's great western mobile belt, is anybody's guess.
In addition to the home page references, this article relies on the sources cited below and in the Colorado Geology Overview article. I'm particularly indebted to Warren Hamilton of the Colorado School of Mines Department of Geophysics for revamping my understanding of plate tectonics through his articles and personal communications.
Unless explicitly attributed to another contributor, all content on this site © Jeremy McCreary. Comments and corrections to firstname.lastname@example.org. Nothing is bought or sold here. Spam will be forwarded to the proper authorities.