Mantle Heterogeneity and Melting Along a Regional Axial Depth Gradient: Th-U Disequilibria Along the Southeast Indian Ridge

Lateral variation in mantle temperature is generally considered to be the main cause for observed global correlations between regionally averaged mid-ocean ridge basalt (MORB) chemistry and ridge axis depth (Klein and Langmuir, 1987). Axial depth should be shallower above hotter mantle because the underlying mantle crosses its solidus at greater depth compared to cooler regions, and leads to greater crustal production. One expectation of this global model is that melting beneath ridges should involve progressively more (deep) garnet peridotite as ridge depth shallows. A negative correlation between (²³⁰Th/²³⁸U) and axial depth in a global dataset (largely dominated by eastern Pacific and Atlantic MORB) generally supports this notion, because garnet is known to fractionate Th from U during melting (Bourdon et al., 1996). The extent to which such global variations reflect only variations in melting conditions of passively upwelling mantle versus other conditions related to mantle heterogeneity or actively fed melting anomalies is presently unknown. To further address this question on the regional scale, we have measured Th-U isotopes on 11 basaltic glasses collected along the Southeast Indian Ridge (SEIR) from ∼90° E to ∼117° E. This ∼2600 km section of ridge is characterized by a west to east gradient in axial depth from ∼2300 m to >4500 m, similar to that of the global ridge system away from the influence of hotspots. However, unlike the global (²³⁰Th/²³⁸U)-axial depth correlation which uses regionally averaged data for ridge segments from a range of spreading rates, the SEIR is spreading at a nearly constant rate of 70-75 mm/yr and is devoid of large transform offsets. These first 11 SEIR glasses were selected to span the geographic range and to be representative of the elemental and isotopic compositions. Th and U concentrations range from 130-940 ppb and 55-267 ppb, respectively, with Th/U values ranging from 2.36 to 3.77. All samples show ²³⁰Th excesses, with (²³⁰Th/²³⁸U) ranging from 1.01 to 1.24. There is no correlation between (²³⁰Th/²³⁸U) and axial depth, in contrast to the simple prediction from the global scale model. On a (²³⁸U/²³²Th)-(²³⁰Th/²³²Th) equiline diagram the SEIR data form 3 geographical, non-collinear groupings: three basalts from the westernmost portion of our study area have the highest (²³⁰Th/²³²Th) and (²³⁸U/²³²Th), the easternmost basalt has significantly lower values, and basalts from a central region (101° E to 114° E) have intermediate values. The 7 central region basalts form a well-correlated positively sloping (²³⁰Th/²³²Th) vs. (²³⁸U/²³²Th) array which is shallower than the equiline and extends from 7% to 24% ²³⁰Th-excess. Overall, (²³⁰Th/²³²Th) shows a strong negative correlation with axial depth and, along with Th/U, correlates well with other isotopic tracers such as ³He/⁴He and ²⁰⁸Pb/²⁰⁶Pb, which vary systematically along axis. In contrast, (²³⁰Th/²³⁸U) shows no systematic variations with these isotopic parameters or axial depth. Unlike the global dataset, the Th-U disequilibria in SEIR MORB suggests that, at a regional scale, melting in the Indian Ocean mantle is primarily controlled by variations in mantle composition.