RATES OF LAVAKA FORMATION IN CENTRAL MADAGASCAR

Madagascar's tectonically oversteepened, deeply weathered central highlands are eroding rapidly. The most visible manifestations of this process are lavakas: deep, vertical-sided groundwater-sapping features that develop in the thick saprolite and laterite of the microcontinent interior. Although widely reported to be of recent vintage and related to human activity, there are in fact few data on their rates of formation. The Miarinarivo area is a prime candidate for increasing erosion rates if lavaka formation is driven by human activity: it is only 100 km from the capital city, served by a paved main road, and the surrounding countryside is well populated and heavily used for grazing and cultivation.

Field work in the Miarinarivo area confirms that individual lavakas grow very rapidly: one lavaka that initiated in October 2002 had grown by August 2004 to two big lobes, each about 50 m long, 12 m wide, and 15 m deep. The volume excavated from this single feature was about 10,000 m³ in 22 months, but the erosion rate decreased through time, from about 1300 m³/month in the first year to about 450 m³/month in the second (integrated across the whole year, although erosion is actually strongly concentrated in the 5-month rainy season).

However, although the lavakas in the Miarinarivo area appear very fresh and recent, most of them are not. Of 240 lavakas in the mapping area in 2004, 227 appear on air photos taken in 1949, which means that the lavaka increase rate in this area is only 0.08%/year, or 0.1/km² yr^-1. Old lavakas are not significantly bigger than they were in 1949, although in many cases their headwalls remain vertical and vegetation-free. We infer that individual lavakas tend to reach a maximum size early on, with minimal growth thereafter. Sediment output declines exponentially, but limited sediment output may be maintained over many decades. The sediment comes mainly from small-scale headwall collapse, probably in the form of fluidised flows and low-density debris flows. Present data do not allow us to evaluate how long it takes for lavakas to become inactive, lose their vertical walls, and relax back to bowl-shaped depressions in the landscape; but cosmogenic isotope studies in progress will shed light on this timescale.