Erosion via lavaka formation is widespread in Madagascar, but controls on why and where lavakas occur are not understood. Geographic information system analysis reveals a spatial correlation between lavaka abundance and the frequency of seismic events: most lavakas occur in or near areas where recorded earthquakes (magnitude 0.5-5.6) are most frequent. This correlation explains the unevenness of lavaka distribution in the Malagasy highlands, and highlights the importance of natural factors in lavaka formation. Seismic activity appears to precondition the landscape to lavaka formation, although the mechanism by which this happens is not yet known. Recognizing the connection, however, allows us to pinpoint areas prone to future lavaka development in zones of active deforestation. Areas with the greatest frequency of seismic events are most at risk for high-density lavaka development. This idea came originally from a conversation with Mike Zavada. Neil Wells kindly shared his fi eld data. Sharron Macklin and David Backus helped with the geographic information systems analysis. Land-Sat images came from the Global Landcover Facility, University of Maryland. Didier Bertil and the Institut et Observatoire de Géophysique d'Antananarivo (IOGA) provided seismic data.

A B S T R A C TThe central highlands of Madagascar are characterized by rolling hills thickly mantled with saprolite and cut in many areas by dramatic gullies known as lavakas. This landscape generates sediment to rivers via diffusive downslope movement of colluvium and event-driven advection of material from active lavakas; these two sediment sources have very different 10 Be signatures. Analyzed lavaka sediment has very little 10 Be ( atoms 10 Be g Ϫ1 ), 5 0.8-10 # 10 consistent with deep excavation liberating previously shielded saprolite with little exposure to cosmic rays. Colluvium, in contrast, has greater 10 Be concentrations ( atoms 10 Be g Ϫ1 ), reflecting long residence times in the 5 6-21 # 10 near-surface environment. Comparison of 10 Be abundance in hillslope, lavaka, and river sediment samples indicates that lavakas dominate the mass input to rivers (84% by volume) in spite of the fact that they occupy a small fraction of the land surface area. River terrace sediments that are at least a millennium old have 10 Be concentrations indistinguishable from those of modern lavaka-dominated river sands, from which we infer that lavakas were widespread on the landscape at or before the time that humans colonized the central highlands. Erosion rates derived from cosmogenic 10 Be in river sediment average approximately 12 m m.yr. Ϫ1 , or about 32 t km Ϫ2 yr Ϫ1 , which is three orders of magnitude lower than commonly reported erosion rates for Madagascar. q1 Figure 1. Topography and erosion in the central highlands of Madagascar. A, Convex-slope topography along Route Nationale 4, north of Antananarivo. The rolling grasslands commonly form convex hillslopes in the lateritized central highlands, at elevations ≈500-2000 m. In many places, remnant riparian forests remain in the declivitous valleys, as shown here; in other areas, the drainages are occupied by rice fields. B, Active stage II (sensu Wells et al. 1991) lavaka in the Amparafaravola region. The headwall is 33 m high, and the total width is 41 m. Sample 2005-2 (table 1) was collected from outfall sediments. C, Lavakas incising demi-orange topography in the Miarinarivo region. The nearest lavaka is the most active (stage II of Wells et al. [1991]), as shown by the steep and bare walls. It is 50 m long and 16 m deep. The lavakas in the background are less active (stage III of Wells et al. [1991]): their walls are not quite so steep, and the muted color is given by a clay veneer that coats noneroding walls. D, Stage II lavaka from the Miarinarivo region. This is the location of sample 2004-4 (table 1). The lavaka is 16 m deep, and the total width is 40 m. E, Ancient lavaka from the Amparafaravola region, similar to the sites of 2004-1, 2004-8, and 2005-5. Although the depression is fully vegetated, with no actively eroding walls, the characteristic lavaka shape is preserved: a deep amphitheater with concave internal slopes (in contrast to the convex slopes of the surrounding demi-orange landscape), a flat floor, and a narrow outlet. A color version of this...

The characteristic gullies of central Madagascar-lavakas-vary greatly in abundance over short distances, but existing understanding does not explain why some hillsides should have high concentrations of lavakas when nearby slopes have fewer. We present a GIS analysis of lavaka abundance in relation to bedrock geology and topography, covering two areas in the central highlands: the region near Anibatondrazaka and that around Tsaratanana. Both regions have similar average lavaka density (6 lavakas/km^ in Ambatondrazaka, and 5 lavakas/km^ in Tsaratanana, but local lavaka concentrations vary widely. Individual one-km^ squares can host up to 50 lavakas/km-in Tsaratanana and up to 150 lavakas/km^ in Ambatondrazaka. We find no predictive relationship between bedrock type and lavaka abundance. There is, however, a relationship between lavakas and slope such that lavakas increase in abundance as slopes get steeper, up to an optimum steepness, beyond which they become le.ss numerous. The optimum steepness for lavaka development is about 10 to 15° in Tsaratanana and 25 to 30° in Ambatondrazaka. Lavakas also seem to favour slopes where the gradient changes locally, with an optimum change in grade somewhere in the range 2 to 5°. Our results provide quantitative constraints on lavaka distribution that can be tested in other areas.

Abstract. Petrographic features, mineralogy, and stable isotopes from two stalagmites, ANJB-2 and MAJ-5, respectively from Anjohibe and Anjokipoty caves, allow distinction of three intervals of the Holocene in NW Madagascar. The Malagasy early Holocene (between ca. 9.8 and 7.8 ka) and late Holocene (after ca. 1.6 ka) intervals (MEHI and MLHI, respectively) record evidence of stalagmite deposition. The Malagasy middle Holocene interval (MMHI, between ca. 7.8 and 1.6 ka) is marked by a depositional hiatus of ca. 6500 years.Deposition of these stalagmites indicates that the two caves were sufficiently supplied with water to allow stalagmite formation. This suggests that the MEHI and MLHI intervals may have been comparatively wet in NW Madagascar. In contrast, the long-term depositional hiatus during the MMHI implies it was relatively drier than the MEHI and the MLHI.The alternating wet-dry-wet conditions during the Holocene may have been linked to the long-term migrations of the Intertropical Convergence Zone (ITCZ). When the ITCZ's mean position is farther south, NW Madagascar experiences wetter conditions, such as during the MEHI and MLHI, and when it moves north, NW Madagascar climate becomes drier, such as during the MMHI. A similar wetdry-wet succession during the Holocene has been reported in neighboring locations, such as southeastern Africa. Beyond these three subdivisions, the records also suggest wet conditions around the cold 8.2 ka event, suggesting a causal relationship. However, additional Southern Hemisphere highresolution data will be needed to confirm this.

We report the discovery of osumilite in ultrahigh‐temperature (UHT) metapelites of the Anosyen domain, southern Madagascar. The gneisses equilibrated at ~930°C/0.6 GPa. Monazite and zircon U–Pb dates record 80 Ma of metamorphism. Monazite compositional trends reflect the transition from prograde to retrograde metamorphism at 550 Ma. Eu anomalies in monazite reflect changes in fO2 relative to quartz–fayalite–magnetite related to the growth and breakdown of spinel. The ratio Gd/Yb in monazite records the growth and breakdown of garnet. High rates of radiogenic heat production were the primary control on metamorphic grade at the regional scale. The short duration of prograde metamorphism in the osumilite gneisses (<29 ± 8 Ma) suggests that a thin mantle lithosphere (<80 km) or advective heating may have also been important in the formation of this high‐T, low‐P terrane.