Various cavities studied in western Niger and South Cameroon show the existence of important karstic phenomena into metagabbros and gneisses. These large-sized caves resulted from generalized dissolution of silicate formations in spite of their low solubility. Karstification is produced by deep hydrous transfer along lithological discontinuities and fracture net works. The existence of such caves has major implications in geomorphology, under either Sahelian and Equatorial climate, and in hydrogeology and water supply, particularly in the Sahel area.
Introduction. – Since a few decades, several karst-like morphologies are described in non-carbonated rocks (sandstones, quartzites, schistes, gneisses…) [Wray, 1997 ; Vicat and Willems, 1998 ; Willems, 2000]. The cave of Guéssédoundou in West Niger seems to be due to a large dissolution of metagabbros. The cave of Mfoula, South Cameroon, attests for the same process in gneisses. This forms proof that big holes may exist deeper in the substratum even of non-carbonated silicate rocks. Their size and number could mainly influence the landscape and the hydrogeology, especially in the Sahelian areas.
Guéssédoundou, a cave into metagabbros in West Niger. – The site of Guéssédoundou is located 70 km south-west of Niamey (fig. 1). The cave is opened at the top of a small hill, inside in NNE-SSW elongated pit (fig. 2 ; pl. I A). The hole, 3 to 4 m deep and 20 m large, has vertical walls and contains numerous sub-metric angular blocks. A cave, a few meters deep, comes out the south wall. Bedrocks consist of metagabbros of the Makalondi greenstone belt, a belt of the Palaeoproterozoic Birimian Formations of the West Africa craton [Pouclet et al., 1990]. The rock has a common granular texture with plagioclases, partly converted in albite and clinozoisite, and pyroxenes pseudomorphosed in actinote and chlorite. It is rather fairly altered. Chemical composition is mafic and poorly alkaline (tabl. I). A weak E-W schistosity generated with the epizonal thermometamorphism. The site depression was created along a N010o shear zone where rocks suffered important fracturation and fluid transfers, as shown by its silification and ferruginisation.
The absence of human activity traces and the disposition of the angular blocks attest that the pit is natural and was due to the collapse of the roof of a vast cavity whose current cave is only the residual prolongation. To the vertical walls of the depression and at the cave entry, pluridecimetric hemispheric hollows are observed (pl. I B). Smooth morphology and position of these hollows sheltered within the depression dismiss the assumptions of formation by mechanical erosion. In return, these features are typical shape of dissolution processes observed into limestone karstic caves. That kind of process must be invoked to explain the opening of the Guéssédoundou cave, in the total lack of desagregation materials. Dissolution of metagabbro occurred during hydrous transfer, which was probably guided by numerous fractures of the shear zone. Additional observations have been done in the Sirba Valley, where similar metabasite rocks constitute the substratum, with sudden sinking of doline-like depressions and evidence of deep cavities by core logging [Willems et al., 1993, 1996]. It is concluded that karstic phenomena may exist even in silica-aluminous rocks of crystalline terrains, such as the greenstones of a Precambrian craton.
Mfoula a cave into gneisses in South Cameroon. – The cave of Mfoula is located 80 km north-east of Yaoundé (fig. 3). It is the second largest cave of Cameroon, more than 5,000 m3, with a large opening in the lower flank of a deep valley (pl. I C). The cavity is about 60 m long, 30 m large and 5 to 12 m high (fig. 4; pl. I D). It is hollowed in orthogneisses belonging to the Pan-African Yaoundé nappe. Rocks exhibit subhorizontal foliation in two superposed lithological facies: the lower part is made of amphibole- and garnet-bearing layered gneisses, and the upper part, of more massive granulitic gneisses. Average composition is silico-aluminous and moderately alkaline (tabl. I). The cave is made of different chambers separated by sub-cylindrical pillars. The ceiling of the main chamber, 6 m in diameter, is dome-shaped with a smooth surface (D, fig. 4). The walls have also a smooth aspect decorated with many hemispherical hollows. The floor is flat according to the rock foliation. They are very few rock debris and detrital fragments and no traces of mechanical erosion and transport. The general inner morphology is amazingly similar to that of a limestone cave. The only way to generate such a cavity is to dissolve the rock by water transfer.
To test the effect of the dissolution process, we analysed a clayey residual sampled in an horizontal fracture of the floor (tabl. I). Alteration begins by plagioclases in producing clay minerals and in disagregating the rock. However, there is no more clay and sand material. That means all the silicate minerals must have been eliminated. Dissolution of silicates is a known process in sandstone and quartzite caves. It may work as well in gneisses. To fasten the chemical action, we may consider an additional microbial chemolitotrophe activity. The activity of bacteria colonies is known in various rocks and depths, mainly in the aquifer [Sinclair and Ghiorse, 1989 ; Stevens and McKinley, 1995]. The formation of the Mfoula cave is summarized as follow (fig. 5). Meteoric water is drained down along sub-vertical fractures and then along horizontal discontinuities of the foliation, particularly in case of lithological variations. Chemical and biological dissolution is working. Lateral transfers linked to the aquifer oscillations caused widening of the caves. Dissolved products are transported by the vertical drains. Regressive erosion of the valley, linked to the epeirogenic upwelling due to the volcano-tectonic activity of the Cameroon Line, makes the cavities come into sight at the valley flanks.
Discussion and conclusion. – The two examples of the Guéssédoundou and Mfoula caves evidence the reality of the karsts in non-carbonated silicated rocks. The karst term is used to design ≫ any features of the classical karst morphology (caves, dolines, lapies…) where dissolution plays the main genetical action ≫ [Willems, 2000]. Our observations indicate that (i) the karst genesis may have occurred into any kind of rocks, and (ii) the cave formation is not directly dependent of the present climate. These facts have major consequences to hydrogeological investigations, especially for water supply in Sahelian and sub-desertic countries. Some measurements of water transfer speed across either sedimentary pelitic strata of the Continental terminal or igneous rocks of the substratum in West Niger [Estèves and Lenoir, 1996 ; Ousmane et al., 1984] proved that supplying of aquifers in these silico-aluminous rocks may be as fast as in a karstic limestone. That means the West Niger substratum is highly invaded by a karstic net and may hidden a lot of discontinuous aquifers. The existence of this karst system can be easily shown by morphological observations, the same that are done in karstic limestone regions (abnormally suspended dry valleys, collapses, dolines…). Clearly, this must be the guide for any search of water, even in desertic areas where limestones are absent.