Rifting, Cenozoic volcanic and tectonic processes control the landscape of Madagascar

Madagascar’s topography is defined by three distinct features: the western remnant escarpment, a central dissected plateau, and the eastern great escarpment. The modern landscape reflects a complex geological history shaped by multiple phases of rifting. The western escarpment dates back approximately 170 Myr, coinciding with Madagascar’s initial separation from Africa. A second phase of rifting, around 90 Myr ago, marked Madagascar’s separation from the Seychelles-India block, leading to the formation of the eastern escarpment. A final phase of landscape evolution resulted from Late Cenozoic volcanic and tectonic extension of Madagascar’s interior, which led to the westward migration of the water divide away from the escarpment. 

Building on this geological context, we constructed a landscape evolution model to understand how these rifting phases and subsequent processes influence Madagascar's topography using the Divide and Capture (DAC) code. We test the first-order topography by generating two phases of rifting, including the formation of rift escarpments and flexural tilting. We assume that rifting thinned the crust, inducing unloading at each margin with flexural uplift and tilting in response. We find that each rifting phase results in the formation of an escarpment with divide-type river profiles, but that westward flexural tilting during the second phase shifts the main divide eastward, accelerating the disintegration of the western escarpment and creating detached landforms and knickzone-type river profiles. 

Next, we investigate how second-order topographic features can be explained by volcanic activity, intraplate extension, and rock erodibility contrasts. In our model, volcanic activity affects the landscape by steadily building up less erosive topographic edifices. This feature is located on the plateau closer to the eastern escarpment, simulating the real-world scenario. The volcanic topographic highs can locally deflect the topographic gradient such that the major divide “jumps” from its original location and becomes locally pinned to the top of the volcanic edifices. We also explored the influence of surface subsidence in the graben due to intraplate extension on the landscape. We kinematically lowered the plateau surface in the specified rectangular “graben” area by assuming the graben’s longitudinal axis is parallel to the major divide. We find that the progressive retreat of the escarpment erodes the nearest flank of the graben, capturing the enclosed basin of the graben and causing the divide to jump to the furthest flank. These processes reshape the escarpment river morphology but remain confined locally to the graben-affected area. Rock erodibility contrast in the plateau basement is modeled by specifying various shapes of vertical blocks composed of more erosion-resistant rock. These blocks are assumed to have the same initial height as their surroundings and are applied at model initialization. During plateau incision, these blocks erode at a slower rate, causing the escarpment retreat to slow down upon encountering them. As a result, they are left behind as remnant escarpments detached from the plateau.