Formation of oceanic crust at slow spreading rates: New constraints from an extinct spreading center I n the Labrador Sea K. E. Louden* Department of Oceanography, Dalhousie University, Halifax, Nova Scotia B3H 4J1, Canada J. C. Oslert Defense Research Establishment Atlantic, P.O. Box 1012, Dartmouth, Nova Scotia B2Y 3Z7, Canada S. P. Srivastava C. E. Keen Geological Survey of Canada-Atlantic, Bedford Institute of Oceanography, P.O. Box 1006, Dartmouth, Nova Scotia B2Y 4A2, Canada ABSTRACT We present a joint analysis of seismic reflection and refraction profiles across an extinct spreading center in the Labrador Sea. The refraction data are used to constrain the crustal and subcrustal velocity structure, and the deep multichannel reflection data are used to determine the nature of the tectonic fabric within the crust. The resulting crustal model shows that major reductions in crustal thickness and velocity across the rift are associated with a broad zone of rotated fault blocks. Enhanced faulting persists over a width of 100 km, creating a deep rift valley and adjacent elevated ridges. The location of crustal-scale faults beneath the sides of the rift valley bound a region of reduced mantle velocity, consistent with its formation by 50/0-10% serpentinization of mantle peridotite. The variation in crustal thickness across the rift indicates a 300/0-40% reduction in the crustal thickness for full spreading rates <20 km/m.y. However, our results suggest that most of the observed crustal thinning is produced by postaccretionary tectonic extension of the crust, during an extended interval when the rate of melt supply did not keep pace with the rate of plate separation. Such a discontinuous process may not be properly represented by theoretical models of crustal formation that assume steady-state viscous flow. BACKGROUND Extinct spreading centers are mid-ocean ridges where the active accretion of oceanic crust has ceased. They offer sites where we can investigate how the formation of oce- anic crust has responded to a reduction and eventual cessation in spreading along a flow line of a single rift segment. This geometry avoids complications in comparing the structures of different ridge segments, which may be affected by variations in magma sup- ply that are unrelated to changes in spread- ing rate (e.g., regional variations in as- thenospheric temperature and local effects produced by ridge crest discontinuities). Of the few extinct spreading centers that have been recognized, one of the best ex- amples is located in the Labrador Sea (Fig. 1). This ridge formed one limb of a triple junction that separated Greenland, North America, and Eurasia during the Late Cretaceous and early Tertiary. It stopped spreading between chron 21 (49 Ma) and *E-mail: louden@ac.dal.ca. 'Present address: SACLANT Undersea Re- search Centre, Viale San Bartolomeo, 400, 19138 La Spezia, Italy. chron 13 (35 Ma), when the Mid-Atlantic Ridge took its present form (Roest and Sri- vastava, 1989). The basement relief of the extinct rift was filled subsequently by thick sediments. The location of the extinct rift axis, however, is clearly indicated by a lin- eated gravity low, which is centered between paired magnetic isochrons (Fig. 1). SEISMIC PROFILES Two independent seismic data sets are used. The first data set is a wide-angle re- fraction profile (line R2), which consists of record sections of air-gun shots recorded at eight positions along a line situated perpen- dicular to the strike of the extinct rift axis (Fig. 1). A detailed analysis of these data by Osler and Louden (1992, 1995) yields a well- constrained velocity-depth model across the rift axis. This model includes a significant reduction in crustal thickness and crustal and mantle velocities within the extinct rift axis. The second data set (Srivastava and Keen, 1995) consists of a multichannel seis- mic reflection profile, TLS-902, which crosses the same rift segment as the refrac- tion profile (Fig. 1). This profile follows a single flow line during the period of oblique rifting following chron 25. The reflection Geology; September 1996; v. 24; no. 9; p. 771-774; 4 figures. data clearly record the existence of large ro- tated fault blocks within the rift axis, also indicating crustal thinning. In this paper, we combine these two data sets to investigate the relationship between the patterns of faulting and changes in crustal and subcrustal velocity. Although the profiles are not coincident, they may be combined, because the ridge structure is pri- marily two dimensional between the pro- files. The free-air gravity anomaly remains essentially the same across the rift axis (Fig. 1), except at the ends of seismic re- fraction line Rl. Likewise, Osler and Louden (1992) showed that the crustal ve- locity structure along the axis of the rift re- mains primarily one dimensional along the central section of line Rl. The superposition of the data sets (Fig. 2) is made along line TLS-902. The velocity versus depth model for line R2 is replaced by the equivalent velocity versus traveltime model, except that the top of acoustic base- ment is made compatible with line TLS-902 and velocities in the lower crust in zone A are increased by km/s, as suggested by gravity modeling, while still satisfying the re- fraction data (Osler, 1993). Linear projec- 771