[1] There is a systematic variation in axial morphology and axial depth along the Southeast Indian Ridge (SEIR) with distance away from the Australian Antarctic Discordance, an area of cold uppermost mantle. Since spreading rate (72-76 mm/yr) and mantle geochemistry appear constant along this portion of the SEIR, the observed variations in axial morphology and axial depth are attributed to a gradient in mantle temperature. In this study, we report results from a multichannel seismic investigation of on-axis crustal structure along this portion of the SEIR. Three distinct forms of ridge crest morphology are found within our study area: axial highs, rifted axial highs, and shallow axial valleys. Axial highs have a shallow ($1500 m below seafloor (bsf)) magma lens and a thin ($300 m) layer 2A along the ridge crest. Rifted axial highs have a deeper ($2100 m bsf) magma lens and thicker ($450 m) layer 2A on-axis. Beneath shallow axial valleys, no magma lens is imaged, and layer 2A is thick ($450 + m). There are step-like transitions in magma lens depth and layer 2A thickness with changes in morphology along the SEIR. The transitions between the different modes of axial morphology and shallow structure are abrupt, suggesting a threshold-type mechanism. Variations in crustal structure along the SEIR appear to be steady state, persisting for at least 1 m.y. Portions of segments in which a magma lens is found are characterized by lower relief abyssal hills on the ridge flank, shallower ridge flank depths, and at the location of along-axis Mantle Bouguer Anomaly (MBA) lows. The long-wavelength variation in ridge morphology along the SEIR from axial high segments to the west to axial valley segments to the east is linked to the regional gradient in mantle temperature. Superimposed on the long-wavelength trend are segment to segment variations that are related to the absolute motion of the SEIR to the northeast which influence mantle melt production and delivery to the ridge.Components: 10,963 words, 9 figures.

[1] The seismic structure of uppermost crust evolves after crustal formation with precipitation of alteration minerals during ridge-flank hydrothermal circulation. However, key parameters of crustal evolution including depth extent and rates of change in crustal properties, and factors contributing to this evolution remain poorly understood. Here, long-offset multichannel seismic data are used to study the evolution of seismic layer 2A and uppermost 2B from 0 to 550 ka at three segments of the intermediate spreading rate Southeast Indian Ridge. The segments differ in on-axis morphology and structure with crustal magma bodies imaged at axial high and rifted high segments P1 and P2, but not at axial valley segment S1 and marked differences in thermal conditions within the upper crust are inferred. One-dimensional travel time modeling of common midpoint supergathers is used to determine the thickness and velocity of layer 2A and velocity of uppermost 2B. At all three segments, layer 2A velocities are higher in 550 ka crust than onaxis (by 7-14%) with the largest increases at segments P1 and P2. Velocities increase more rapidly (by 125 ka) at P1 with spatial variations in velocity gradients linked to location of the underlying crustal magma body. We attribute these differences in crustal evolution to higher rates of fluid flow and temperatures of reaction at these ridge segments where crustal magma bodies are present. Layer 2A thickens off-axis at segments P1 and P2 but not at S1; both off-axis volcanic thickening and downward propagation of a cracking front linked to the vigor of axial hydrothermal activity could contribute to these differences. In zero-age crust, layer 2B velocities are significantly lower at segments P1 and P2 than S1 (5.0, 5.4, and 5.8 km/s respectively), whereas similar velocities are measured off-axis at all segments (5.7-5.9 km/s). Lower on-axis 2B velocities at segments P1 and P2 can be partly attributed to thinner layer 2A, with lower overburden pressures leading to higher porosities in shallowest 2B. However, other factors must also contribute. Likely candidates include subaxial deformation due to magmatic processes and enhanced cracking with axial hydrothermal activity at these segments with crustal magma bodies.Components: 11,712 words, 10 figures, 2 tables.Keywords: Southeast Indian Ridge; crustal evolution; layer 2B; multichannel seismic; layer 2A.

S U M M A R YAxial and ridge flank depths on the Southeast Indian Ridge (SEIR) systematically increase for 2500 km from 90 • E to 120 • E approaching the Australian-Antarctic Discordance. The SEIR also experiences an abrupt change in ridge axis morphology near 103 • 30 E with axial highs found to the west and axial valleys to the east. Since the spreading rate is constant throughout this region, these variations have been ascribed to an along-axis gradient in mantle temperature. A seismic refraction experiment provides information on the crustal thickness and seismic velocity structure of two segments with differing axial morphology. Segment P2, centred near 102 • E with an axial high has a mean crustal thickness of 5.9 ± 0.2 km, while the mean crustal thickness is 5.3 ± 0.3 km at Segment S1 with an axial valley and centred near 109 • 45 E. Isostatic compensation of the difference in crustal thickness and density structure between the two segments only accounts for 33 m of the 198 m difference in average ridge flank depth between the two segments. The remaining depth difference must result from a difference in mantle density. Melt production models imply a mantle temperature difference of 11-13.5 • C to produce the observed difference in crustal thickness. Isostatic compensation of the two segments requires that the resulting density difference must extend to about 300 km in the mantle.The transition in axial morphology along the SEIR is very abrupt occurring over a narrow zone within a single segment in which the transition is complete. If a linear mantle temperature gradient is assumed, the temperature difference across the transition segment is only 2.4 • C. The change in axial morphology is accompanied by abrupt changes in other parameters including abyssal hill height, magnetic anomaly amplitude, layer 2a thickness and the presence or absence of an axial magma lens. The abrupt, coincident change in a number of parameters with a very small change in mantle temperature strongly suggests a threshold change between two distinctly different modes of crustal accretion. The trigger for the transition appears to be whether a steady-state crustal magma lens can be maintained.