The Mid-Ocean Ridge
 

It is the longest mountain chain, the most active volcanic area and until recently, the least accessible region on the earth. New maps reveal striking details of how segments of the Ridge form and evolve.
  Kenneth C. Macdonald and Paul J. Fox


  On July 8, 1982, we boarded the research vessel Thomas Washington to survey the East Pacific Rise, a volcanic mountain chain that lies under the Pacific Ocean. The Rise is part of the 75,000-kilometer long formation known as the Mid-Ocean Ridge. Like the seam of a baseball, the Ridge winds around the globe from the Arctic Ocean to the Atlantic Ocean, around Africa, Asia and Australia, under the Pacific Ocean and to the west coast of North America. Even though the Ridge is by far the longest structure on the earth, less was known about its features than about the craters on the dark side of the moon.
     Our colleagues from the Scripps Institution of Oceanography had recently equipped the Thomas Washington with a new type of sonar system, made by the General Instrument Corporation. Called SeaBeam, it could map a two-kilometer swath of the ocean floor in a single ping of the sonar. It would, we hoped, reveal the ocean floor in unprecedented detail, providing new insights into the forces that form and shape the Mid-Ocean Ridge.
     After cruising southeast 2,500 kilometers from the Scripps marine facility in San Diego, we intersected the crest of the East Pacific Rise, located at a depth of about 2.5 kilometers. The Rise marks the boundary between the Pacific and Cocos tectonic plates, each a slab of the earth's crust and upper mantle. The plates separate at a rate of about 120 millimeters per year (twice the rate at which a fingernail grows). As the plates move apart, cracks form along the crest of the rise, allowing molten rock to seep up from the mantle. Some of the molten rock overflows onto the ocean floor in tremendous eruptions. The magma then solidifies to form many square kilometers of new oceanic crust each year. Only a few kilometers above this activity, we felt like Lilliputians crawling along the spine of a slumbering giant that might awaken at any time.
     As the SeaBeam probed the spine of this giant, we watched images of the seascape appear on monitors on board the Thomas Washington. We saw some familiar features: the elevated terrain that defines the axis of the ridge and large breaks, called transform faults, that offset ridge segments by hundreds of kilometers. Yet we and Peter F. Lonsdale of Scripps also observed several unknown structures: segments that bend, ridges that overlap and oceanic crust that is warped and distorted near these features.
     Since the early 1980's our colleagues in France, the U.K. and the U.S. have also surveyed many stretches of the East Pacific Rise as well as other parts of the Mid-Ocean Ridge. These efforts have revealed that the Ridge has many lateral discontinuities that partition its axis into segments. Although discontinuities differ in form and behavior, most of them are deeper and less active volcanically than the segments they define. As a result, the crest of the Ridge undulates up and down by hundreds of meters over distances of from 10 to 1,000 kilometers. During the past several years we have come to understand how these discontinuities and segments evolve and how they are related to processes deep in the earth's crust and mantle.


Sea-floor Spreading
American oceanographer Bruce C. Heezen aptly described the Mid-Ocean Ridge as "the wound that never heals". In 1956 he and Maurice Ewing noticed that the earthquakes in the ocean basin define a continuous belt encircling the world. Because the belt coincided with portions of the Mid-Ocean Ridge that were known at the time, they proposed that the earth was girdled by a continuous system of ocean ridges. Ever since their discovery oceanographers and geologists have tried to get a closer look at the Mid-Ocean Ridge to understand its origins.
     The global geologic processes that form and shape the Ridge were not understood until 1960 when Harry H. Hess of Princeton University introduced the concept of sea-floor spreading. Other workers further refined and developed his idea into the theory of plate tectonics. The theory posits that the crust and upper mantle are divided into a few dozen plates, such as the Pacific and the Cocos, which can move with respect to one another. If two plates separate, material from the mantle can well up, forming a ridge and new oceanic crust.
     The theory of plate tectonics accounts for the largest structures of the Mid-Ocean Ridge. Yet as early as 1960 H. William Menard of Scripps and Heezen discovered that the Mid-Ocean Ridge is a discontinuous structure. As they mapped the Ridge with sounding devices, they found several places where it was offset at right angles to its length. In 1965 J. Tuzo Wilson of the University of Toronto identified these discontinuities as transform faults: a boundary formed perpendicular to the length of the ridge, where the edges of tectonic plates slide past each other in opposite directions. Later Richard N. Hey of the University of Hawaii realized the segments defined by two transform faults could shift in a direction parallel to the length of the ridge. This type of discontinuity was called a propagating rift.
     By the 1980's oceanographers had identified many transform faults and propagating rifts. They had also determined that different parts of the Mid-Ocean Ridge evolved at rates. On the one hand, the plates forming the East Pacific Rise separate at "fast" rates of from 60 to 170 millimeters per year [see "The Crest of the East Pacific Rise," by Kenneth C. Macdonald and Bruce P. Luyendyk; SCIENTIFIC AMERICAN, May 1981]. On the other hand, the plates forming the Mid-Atlantic Ridge are pulled apart at slower rates of about 30 millimeters per year. Because of variations in spreading rates and the rate at which magma is supplied to ridges, the topography of fast-spreading ridges differed from that of slow-spreading ones. The crest of a fast-spreading ridge is defined by an elevation of the oceanic crust several hundred meters high and five to 20 kilometers wide. In contrast, the axis of a slow-spreading ridge is characterized by a rift valley a few kilometers deep and about 20 to 30 kilometers wide.
     In the early 1980's, based on observations of the Mid-Atlantic Ridge (Figure 1), Hans Schouten of the Woods Hole Oceanographic Institution, Kim D. Klitgord of the U.S. Geological Survey and their co-workers proposed that transform faults partition the Mid-Ocean Ridge into segments that behave independently of one another. They suggested that each spreading segment was connected to a source region deep in the mantle. Their data indicated that each segment is about 50 kilometers long and that the source regions an their associated ridge segments remain unchanged for tens of millions of years.


Figure 1.

When we and Lonsdale mapped large portions of the axis of the East Pacific Rise in 1982, we expected the Rise would be a fairly continuous structure, because only nine widely spaced transform faults had been discovered along its 5,000-kilometer length. To our surprise, the axis of the Rise was frequently disrupted by many small offsets (more than 40 have been mapped to date). These discontinuities partitioned the Ridge into segments ranging in length from 10 to 2000 kilometers. Unlike transform faults, these offsets were characterized by overlapping ridge tips, and they did not have a clearly defined fault that connected the tips [Figure 2]. Since their discovery we have mapped the off-axis regions around these overlapping offsets and have learned that the features evolve rapidly. In addition, we have found that the discontinuities can migrate along the Ridge, at varying speeds and in various directions. Individual segments bounded by those discontinuities can apparently lengthen or shorten. High-resolution maps have also documented similar non rigid discontinuities on the slow-spreading Mid-Atlantic Ridge.



Figure 2.


Magma-Supply Model
To determine the origin of these discontinuities, we and our colleagues attempted to find connections between segmentation and volcanic activity. Although volcanism can change greatly from one segment to the next, it does vary systematically along the length of each segment. The least active regions are deep discontinuities, whereas the most active regions are shallow centers of segments [see "The Oceanic Crust," by Jean Francheteau; SCIENTIFIC AMERICAN, September, 1983].
From these observations and others, we, Schouten and our colleagues developed a magma-supply model of ridge segmentation. In the mantle at a depth of from 30 to 60 kilometers, rocks are heated to high temperatures, but because they are usually subjected to high pressure, they remain in a solid state. The environment is somewhat different at the boundary between tectonic plates. As plates separate, some of the rock decompresses and melts. The molten rock then percolates up through the mantle and fills a shallow chamber in the crust beneath the crest of the ridge. As the chamber swells with magma and begins to expand, the crest of the ridge can be pushed upward by the buoyant forces from both the molten rock in the magma chamber and the border region of hot rock in the upper mantle [Figure 3].


Figure 3.

According to the magma-supply model of segmentation, the greater the supply of molten and hot rock to a region, the higher the overlying ridge segment will be elevated. Furthermore, the rate and volume of the molten rock supply can change from region to region, creating variations in the morphology of the different overlying segments.
     The magma-supply model also accounts for smaller structural variations. As magma in the chambers migrates laterally along the ridge axis, the thin, brittle crust above the magma chamber stretches and fractures. The magma can erupt through these fractures to the ocean floor. As the cracks continue to grow, volcanic eruptions flow in their wake. The eruptions will continue until the production of magma subsides and the supply of magma is exhausted. Temporal variations in melt delivery affect a segment's evolution: when a segment is well supplied with molten rock as compared with its neighbors, the segment tends to lengthen, and when it is poorly supplied, the segment shortens. It is this swelling and shrinking of the magma-supply system, in response to plate separation, that initiates the lengthening or shortening of segments and the migration of small discontinuities.
The magma supply model appears to agree with seismic and gravitational measurements of the East Pacific Rise. Seismic measurements reveal that a good reflector of sound energy exists about 1.2 to 2.5 kilometers beneath the shallow portions of each ridge segment. This reflector often deepens and then disappears near discontinuities. In 1987 Robert S. Detrick of the University of Rhode Island and his co-workers proposed that the reflector is the roof of a magma chamber. The strength of sound reflection can be explained by a thin cap of nearly 100 percent melt across the top of the chamber.
Most geologists and oceanographer now agree the reflector is a long, shallow body of magma beneath the ridge surrounded by hot rock. John A. Orcutt of Scripps and his colleagues have made seismic measurements along the northern East Pacific Rise, which suggest such a chamber of molten rock is only two to four kilometers wide and less than one kilometer thick. The magma chamber is surrounded by a wider region of very hot (perhaps slightly molten) rock. The reservoir may be six to 10 kilometers wide and three to six kilometers thick. This region of hot rock extends at least to the base of the oceanic crust and probably a few kilometers into the upper mantle [Figure 4].


Figure 4.

The presence of magma chambers and hot-rock reservoirs has been supported by precise measurements of the gravitational field there, which indicate the presence of a buoyant mass beneath the ridge axis. From both seismic and gravitational measurements, workers have deduced that the magma chamber resembles a mushroom in cross-section: it has a narrow stalk of partial melt feeding a wide buy very thin lens of pure melt.
     Seismic evidence has not definitively proved that magma chambers exist beneath slow-spreading formations such as the Mid-Atlantic Ridge. Other measurements, however, seem to support a magma-supply model for slow-spreading ridges. Donald W. Forsyth and Ban-Yuan Kuo of Brown University and Jian Lin and G. Michael Purdy of Woods Hole found anomalies in the gravitational field, which were centered over the shallowest portions of several segments of the Mid-Atlantic Ridge. The best explanation for these anomalies is an upwelling of hot mantle material or a thickening of the ocean crust beneath the shallow portions of each segment. Both interpretations are consistent with the magma supply model.
     It was a great relief that the seismic and gravitational measurements supported, at least in a general sense, the magma-supply model of segmentation. We and many other tectonicists and geochemists had stuck our necks out fairly far with that hypothesis. True, some of us thought the chamber would be larger, and it remains to be seen if magma actually flows laterally below the ridge axis, but significant evidence has been found to support the model.


Discontinuities
The magma-supply model has been quite successful in accounting for the many different types of discontinuities and segments. Such structures are classified as first, second, third or forth order according to their size, longevity, geometry and behavior. It has been that first-, second-, and third-order structures are fundamental components of both fast- and slow-spreading ridges. (The role of fourth-order features remains unsolved.) Because these structures have been investigated in more detail on fast-spreading ridges, we will describe them in that setting first.
     The most common type of first-order discontinuity is the transform fault. It appears where rigid plates slide past each other. First-order discontinuities offset the ridge segments by at least 20 kilometers and usually more than 50 kilometers. hence, most transform faults were large enough to be revealed by early reconnaissance-mapping efforts. These discontinuities typically define segments from 200 to 800 kilometers long.
     On the ocean floor, transform faults appear to be narrow, straight bands linking the ends of segments. These bands can be traced in the flanks of a ridge for hundreds to thousands of kilometers [Figure 5]. Such traces indicate that first-order structures persist for millions to tens of millions of years.


Figure 5.

A first-order segment can be broken up by several second-order discontinuities that are usually spaced from 50 to 300 kilometers apart. Unlike first-order structures, however, second-order segments are not rigid, and their motion is not concentrated along a narrow fault zone. Second-order discontinuities are complex features characterized by oblique and overlapping structures.
     Second-order discontinuities are typically features that resemble the arms of two people who are preparing to shake hands. The arms (ridges) are extended in such a way that the hands (the curved ends of ridges) overlap. The distance between the "hands" varies from one to 20 kilometers. The offset is typically three times shorter than the distance that the ridges overlap. Such a feature is known as an overlapping spreading center [Figure 6].


Figure 6.

When overlapping spreading centers were discovered in 1982, we could not account for many of their characteristics. Why did so many centers have an overlap-to-offset ratio of 3 to 1? What happened to the crust that lies between the overlapping ridges? Why did the ridges create a distinctive curving pattern?
     In 1984 David D. Pollard of Stanford University, Jean-Christophe Sempere, then at the University of California, Santa Barbara, and one of us (Macdonald) found that the highly repetitive shape of overlapping spreading centers could be explained by the way cracks develop and propagate along ridges. As tectonic plates are pulled apart, cracks form perpendicular to the direction of tension. In the middle of a segment the direction of stress is usually perpendicular to the ridge axis, so the cracks will lengthen parallel to the ridge. In the region of overlapping segments, however, the direction of stress can vary. As a crack from the middle of a segment begins to grow toward the region of overlap, the crack first deflects away from the region and then hooks towards it [Figure 7].


Figure 7.

The crack allows magma to erupt onto the ocean floor, and a new ridge tip is formed. But once the cracks overlap by a distance that approaches three times their offset, the crack propagation stalls abruptly. Soon after, a new crack begins to develop behind the first. As the second develops, the first ridge tip is shed off onto the flanks because of plate separation.

Off-Axis Structures
Spreading centers that overlap by more than several kilometers usually leave "wakes" of deformed oceanic crust up to 80 kilometers wide. The ocean floor within such a disturbed region, called a discordant zone, is 100 to 300 meters deeper than the shallow, magmatically robust portions of the ridge segments. These features have emerged from maps that several expeditions have made of the flanks of the East Pacific Rise. The maps of the discordant zones also show curved fossil ridge tips 10 to 40 kilometers long, which have been cut off at overlapping spreading centers.
     The magma-supply model appears to account for the structure of the overlapping spreading centers. It seems that overlapping spreading centers are at the ends of magma sources and tend to be deprived of magma. If this is true, the crust created at overlapping spreading centers may be up to 50 percent thinner than the six-kilometer-thick crust near the centers of each segment. Detailed seismic and gravitational measurements need to be made in these areas to test this idea.
     Measurements of the earth's magnetic field at overlapping spreading centers support the idea that such centers occur where the magma supply is low. It turns out that lava that erupts from small magma chambers, which alternately solidify and become replenished, tends to contain more iron-rich minerals in a highly magnetized state. On the other hand, magma chambers large enough to remain molten between episodes of magma replenishment produce lava that is magnetically weak. Because rock near overlapping spreading centers is often much more strongly magnetic than rock elsewhere along the ridge, it seems likely that the centers are fed discontinuously from isolated pockets of magma.
     Based on the age of the crust into which the discordant zones extend and on the patterns of off-axis wakes, Laura J. Perram, Suzanne M. Carbotte and Marie-Helene Cormier of the University of California at Santa Barbara have demonstrated that second-order segments persist as discrete entities for up to several million years. The discontinuities may slowly oscillate in position by 10 to 20 kilometers on the ridge or may migrate along the ridge many tens of kilometers at rates of 20 to 100 millimeters per years. A discontinuity tends to move in spurts; a ridge segment can lengthen at rates of several hundred millimeters per year but then may retreat and shorten for a time before making the next surge forward. In this way, the ridge tips at a second-order discontinuity appear to be "dueling" as they surge back and forth along the ridge, generally making slow progress in either direction [Figure 7].


Small Overlaps and deval's

Along the East Pacific Rise, third-order discontinuities usually consist of overlapping spreading centers that offset the ridge by less than three kilometers. Segments defined by third-order discontinuities are from 30 to 100 kilometers long. Third-order discontinuities have been shown to correspond with breaks in magma chambers.
The ridge segments defined by third-order discontinuities leave little or no evidence of off-axis wakes. Because they produce little trace in old oceanic crust on the ridge flanks, we can conclude that third-order discontinuities are geologically short-lived. In fact, we estimate that they are younger than 10,000 years-the time to takes a fast-spreading ridge to generate two kilometers of crust.
     Fourth-order discontinuities are either subtle bends or tiny offsets less than 500 meters in size. The structures are often called deval's (for slight DEViations in Axial Linearity). Deval's are usually spaced from 10 to 40 kilometers apart. A deval may be the manifestation of a single major eruption and therefore may be as young as hundreds to thousands of years old.
     Deval's are very difficult to detect. They can barely be resolved with sonar systems such as SeaBeam, and seismic measurements are not much help either. In some cases the magma chamber beneath a deval deepens slightly and, in rare instances, exhibits an apparent break. In most cases the chambers below fourth-order discontinuities are fairly continuous. During the 1982 cruise one of us (Fox) pointed out to the other (Macdonald) that he had found several deval's in the SeaBeam maps. Macdonald then told Fox that he had been staring at the maps too closely on a rolling ship. We soon agreed that we should focus on the larger offsets if we wanted people to believe our ideas.
Indeed, fourth-order segments (the sections of ridge between deval's) were not recognized as distinct and significant features until 1986, when Charles H. Langmuir of the Lamont-Doherty Geological Observatory, John F. Bender of the University of North Carolina at Charlotte and their colleagues analyzed the geochemistry of a 500-kilometer stretch of the East Pacific Rise. The workers collected rock samples from precise sites on the ocean floor to see if structural segmentation could be associated with variations in rock chemistry. They found that the rocks within each fourth-order segment had a similar composition, whereas rocks from other segments had different chemical signatures. Such measurements have documented the fundamental nature of segmentation over a range of scales and have helped to identify many other deval's.
Why do deval's differ in chemical composition? One theory posits that small blockages divide the magma chamber beneath adjacent fourth-order segments. These divisions would prevent the mixing of the magma in the chambers. Yet only a few such breaks have been detected in magma chambers beneath deval's. Another theory suggests that small batches of molten rock from isolated sources in the upper mantle may be injected locally into a magma chamber and may erupt onto the ocean floor before much mixing occurs. This process would create a fourth-order segment with a distinctive rock chemistry. More evidence is needed to confirm or refute these ideas.
We, Carbotte and Nancy R. Grindlay of the University of Rhode Island have documented several kinds of first-, second- and third-order discontinuities at slow-spreading ridges in the South Atlantic. Like the first-order discontinuities on fast-spreading ridges, the slow-spreading counterparts are transform faults. Second-order discontinuities on slow-spreading ridges are defined by a lateral step of the rift valley or by a deep oblique basin along which the offset rift valleys are linked. The second-order discontinuities persist for millions of years-longer on average than second-order discontinuities on fast-spreading centers. Second-order discontinuities on slow-spreading centers also migrate more slowly along the ridge axis than their fast-spreading counterparts. Third-order discontinuities at slow-spreading centers are small offsets in long volcanic chains within the rift valley floor, whereas fourth-order discontinuities may be small gaps between volcanoes.


Aquatic Life and Segmentation
Oceanographers, tectonicists and geochemists have just begun to understand some of the implications of segmentation for both slow- and fast-spreading ridges. We have found clear examples of first-, second-, third- and fourth-order structures and everything in between. Do segments evolve from fourth through first order and back again? We know that segmentation has been a fundamental process for at least 100 million years. Has segmentation played a role over a much longer period? Investigators have studied exotic faunal communities that flourish near hot springs on the Mid-Ocean Ridge. Can the survival and migration of these communities be linked to the longevity of a given segment?
     These questions will be the focus of research for a program called the Ridge Interdisciplinary Global Experiments (ridge). Among the many goals of the program are to map the axis and flanks of the entire Mid-Ocean Ridge and to generate more detailed images of off-axis features. Even today geologists and oceanographers have mapped less than 5 percent of the sea floor. More than half the earth's crust remains to be explored.