or collective redistirbution of any portion of this article by photocopy machine, reposting, or other means is permitted only with the approval of Th e Oceanography Society. Send all correspondence to: [email protected] or Th eOceanography PO Box 1931, Rockville, USA.Society, MD 20849-1931, or e to: [email protected] Oceanography correspondence all Society. Send of Th approval portionthe ofwith any articlepermitted only photocopy by is of machine, reposting, this means or collective or other redistirbution articleis has been published in Th SPECIAL ISSUE FEATURE Oceanography Ultraslow-Spreading Ridges Threproduction, Republication, systemmatic research. for this and teaching article copy to use in reserved. e is rights granted All OceanographyPermission Society. by 2007 eCopyright Oceanography Society. journal of Th 20, Number 1, a quarterly , Volume Rapid Paradigm Changes BY JONATHAN E. SNOW AND HENRIETTA N. EDMONDS Ultraslow-spreading ridges (< 20 mm yr-1 full rate) represent largely on their spreading rate: “fast” (> 60 mm yr-1 full rate) a major departure from the style of crustal accretion seen in and “slow” (< 60 mm yr-1 full rate). An “intermediate” type is the rest of the ocean basins. Since the 1960s, observations of often placed between them. Both types of ridges share certain fast- and slow-spreading mid-ocean ridges of the Pacifi c and characteristics: (1) they have roughly the same crustal thickness Atlantic Oceans, combined with those of ophiolites (pieces of (6–7 km—see Figure 2), (2) in plan view they have a charac- oceanic crust that have been thrust onto land through plate- teristic stair-step geometry of volcanic rifts separated by per- tectonic processes), were used to defi ne the conceptual struc- pendicular transform offsets, and (3) they generate a charac- tural, tectonic, and hydrothermal architecture of oceanic crust. teristic outcrop pattern of elongate, fault-bounded abyssal hills Over the last 15 years, studies of ultraslow-spreading ridges trending normal to the spreading direction. Their differences have identifi ed several anomalies that cannot be explained lie primarily in their across-axis morphology: fast-spreading by the standard model of oceanic crustal formation. Thus, in ridges have an axial rise with a very narrow summit graben that recent years, ultraslow-spreading ridges have become recog- is the locus for most volcanic and tectonic activity, whereas nized as a class unto themselves. Their “anomalous” charac- slow-spreading ridges have rugged rift mountains enclosing a teristics in fact provide key information about many of the broad axial valley. Fast-spreading ridges tend to be dominated underlying processes that govern crustal accretion at all spread- by volcanism, while the morphology of slow-spreading ridges ing rates. Ultraslow ridges include the Southwest Indian Ridge is dominated more by tectonics. This distinction was widely (between Africa and Antarctica), the Gakkel Ridge (which accepted for many years as a basic principle by the worldwide bisects the Arctic Ocean), and several smaller ridges (Figure 1). geologic community. It was taught to undergraduates as a fun- Since the advent of plate-tectonic theory, mid-ocean ridges damental characteristic of seafl oor geology, and its causes were have been classifi ed based on their structural, morphological, debated as one of the not-completely understood corners of and volcanologic characteristics into two major types based plate-tectonic theory. 90 Oceanography Vol. 20, No. 1 Figure 1. Global plate boundaries (gray) and oce- LT GR anic spreading ridge segments as defi ned KR by Bird (2003). Green indicates fast- MR spreading ridges (full spreading rate ≥ 60 mm yr-1). Red indicates ultraslow-spreading ridges. All other ridge segments are indicated in yellow. GR = Gakkel Ridge, LT = Lena Trough, CT KR = Knipovich Ridge, MR = Mohns Ridge, BY JONATHAN E. SNOW AND HENRIETTA N. EDMONDS CT = Cayman Trough, AAR = America- Antarctic Ridge, and SWIR = Southwest Indian Ridge. Note that sections of the Kolbeinsey and Reykjanes Ridges SWIR north and south of Iceland, which spread at < 20 mm yr-1, AAR are not indicated in red because the infl uence of the Iceland hotspot causes these ridges to behave diff erently than other ultraslow-spreading ridges. 10 9 “Normal” Oceanic Crust 8 Figure 2. Crustal thickness, determined by seismic 7 refraction, as a function of spreading rate. Normal oce- 6 anic crust has a mean thick- ness of 6 km at all spreading 5 rates above 20 mm yr-1. Modifi ed after Bown and ustal ickness (km) 4 White (1994) 3 ismic Cr ismic Se 2 1 Ultraslow Cutoff 20 mm/yr 0 0 20 40 60 80 100 120 140 160 Full Spreading Rate (mm/yr) Oceanography March 2007 91 The characteristics of both slow- and gabbro, and mantle (see Figure 3) in a PROBLEMS WITH THE fast-spreading ridges fit well with the thickness and proportion consistent with PENROSE MODELS ophiolite model for the formation of the seismic structure of both fast- and Outcrop of mantle ultramafic rocks on oceanic crust, which entered the geologi- slow-spreading crust. These characteris- the ocean floor was first described at cal canon at the 1972 Penrose Confer- tics have been an important cornerstone the slow-spreading Mid-Atlantic Ridge ence on Ophiolites and Ocean Crust of plate-tectonic theory for the past (Aumento and Loubat, 1970). In the (Conference Participants, 1972). This 35 years and have continuously proven Penrose model, a 6-km layer of basaltic model, based on geologic mapping on useful in helping to understand the rock covers the mantle; thus, ultramafic land in ophiolites, calls for a layered most inaccessible parts of Earth’s crust rocks at the seafloor should be rare. Their structure of pillow basalts, sheeted dikes, (Nicolas, 1995). emplacement to the ocean floor requires Figure 3. Models of oceanic crustal structure (Nicolas, 1995). The harzburgite-type model describes crust similar to “normal” Penrose-style mid-ocean ridge crust. The origin of lherzolite-type crust was debated for many years, but is now correlated with nonvolcanic rifted margins and to ultraslow- spreading ridges. 92 Oceanography Vol. 20, No. 1 mechanisms that would seem implau- (Cannat et al., 1999; le Roex et al., 1992; ness across the entire range of spread- sible, such as faults with a minimum of Patriat et al., 1997) ing rates (see Figure 2) began to unravel 6-km displacement (though such faults Another major change to ideas of as well. While it remained true at most were in fact later found), or serpentinite crustal accretion came through discov- spreading ridges that the seismically diapirism when the serpentinites them- eries at ridge-transform intersections. determined crustal thickness was nearly selves have densities hardly less than the Dredging results showed that the elevat- constant, at the slowest spreading rates, basaltic rocks through which they must ed inside-corner-high sections of trans- notably in a seismic study done through rise. Also, there is no indication how form faults contained abundant rocks the ice of the Arctic Ocean, the oceanic water might penetrate through many from the lower crust and upper mantle. crust seemed to be dramatically thinner kilometers of oceanic crust to create ser- While the faults bounding the inside than along the rest of the global mid- pentinites in the mantle. Early on, these problems were explained by the idea that great transform faults (Morgan, 1968), which offset ridge segments, provided a pathway for water to enter the mantle Ultraslow-spreading ridges have and for serpentine diapirs to rise to the recently emerged as a unique, new class surface. Most abyssal peridotites were of mid-ocean ridge spreading centers. recovered from the walls of large-offset transform faults, or near them. On the Southwest Indian Ridge (SWIR), peri- dotites were even more common (Dick, 1989; Engel and Fisher, 1975). corner high were large, with as much ocean ridge system (Figure 2). The crust By the mid-1990s, unusual ridge seg- as 500 m of obvious vertical displace- was so thin, in fact, that the very concept ments were also found on the SWIR that ment, this was nowhere near enough to of a “crust” had to be called into ques- were anomalously deep and not per- bring up mantle rocks from beneath a tion, as the seismic structures found pendicular to the spreading direction, full section of oceanic crust (Dick, 1989; could easily be satisfied by a thin layer as Penrose-style volcanic rifts are sup- Karson and Dick, 1983). Subsequently, of serpentinite overlying bare mantle posed to be. Neither were they transform improved bathymetric imaging showed (Bown and White, 1994; Jackson et al., faults, which are geometrically required obvious signs that this faulting was along 1982; Reid and Jackson, 1981). to be parallel to the spreading direction. many tens of kilometers of displace- The walls of these rifts instead trend at ment (Cann et al., 1997; Tucholke et al., EvIDENCE FROM THE FROZEN a highly oblique angle to the prevail- 1998). Nearby dredging of anomalously NORTH: AMORE 2001 ing spreading direction. These oblique deep regions of the ridge (so-called zero- A consensus thus evolved during the rift segments frequently contain mantle offset fracture zones, or nontransform 1990s that while the Penrose model still peridotites as well as basalts of unusual discontinuities) also recovered lower held for most mid-ocean ridge systems, composition. Although these types of crustal and upper mantle rocks, far from basalt were already known from ocean the effects of transforms (Cannat et al., JOnaTHan E. SnOW ([email protected]) is islands and ridges near major hotspots, 1995). At this point, one was forced to Assistant Professor, Department of Geo- their presence in these anomalous rifts, wonder, even on slow-spreading ridges, sciences, University of Houston, Houston, often called “leaky transforms,” could just how much of the crust could be TX, USA.