GEOMORPHOLOGICAL APPROACH TO THE AIT ABDI KARST PLATEAU
(CENTRAL HIGH ATLAS, MOROCCO)

Luc Perritaz & Michel Monbaron
Institute of Geography, Faculty of Sciences, University of Fribourg, 
PŽrolles, CH-1700 Fribourg (Switzerland)

Introduction
A. Geographic situation
The Ait Abdi karst plateau is located at the core of the calcareous High Atlas (32 N/6W, Fig. 1). It has an area of 160 km2. It ranges between 2,200 and 3,000 meters a.s.l. and it is located 800 to 1000 meters above the bottom of the surrounding valleys. 
This region has been the object of few regional studies which have mainly dealt with geologic (geological maps by Jossen, 1990 and Milhi, to be published), geomorphological (Couvreur 1974 and 1988, Monbaron & Perritaz 1993, Perritaz 1995, 1996), hydrogeological (Monbaron & Perritaz 1994, Perritaz 1995) or speleological aspects.
B. Geologic context: lithology, structure and tectonics
The plateau consists of a thick complex of massive Bajocian limestone (Jossen, 1990) which form a vast syncline gently dipping to NE. These limestones, which are 300 meters thick, overlie another thick series of Toarcian-Aalenian clastic sediments forming the regional aquiclude (Fig. 2). The plateau is limited to the N and to the S by a sharp change in the dip of sedimentary layers (ejective thrusted anticlines), and to the W and E by deep canyons created by large rivers. It follows that the plateau is a perfectly isolated limestone compartment both from a morphologic and from a hydrogeologic point of view. It can be considered as a closed system except for the northern part which is much less known.
We found many large vertical doleritic dykes and faults. Four groups of these were determined on the base of direction and length statistics: N000-010, N040-060, N080-090 and N100-120. The dykes, which are very important for locations of karst depressions, seem to function as major drainpipes for groundwater. In fact, many springs correspond to outcrops of these dykes.
C. Climatic context
The climate of this area is of Mediterranean type influenced by altitude: maximum rainfall during the winter and spring seasons; not lasting even if relevant snow covering; the dry summer season characterised by storms; rainfall ranging from 500 to 700 mm/year (out of which 60% of snow), and actual evapotranspiration of approximately 400 mm/year. All this shows that the aquifer recharge is limited, but that the large barren plateau surfaces which present typical and well developed karst forms (dolines, poljes, dry valleys, shafts...) improve the infiltration rate. The infiltration occurs only between December and April while, during the dry season, potential evapotranspiration is too high to permit infiltration (Fig.3). Since the runoff is close to 0, we can calculate the approximate amount of annual infiltration:
650 mm rainfall- 400 mm effective evapotranspiration = 250 mm effective infiltration
Thus, for the 160 km2 plateau, there is 40.8 per 106 m3 of annual aquifer recharge, i.e. 1.3 m3/sec mean discharge for all springs combined. Unfortunately, we cannot verify this amount by means of continuous measurements during spring time, and consequently we cannot establish a valid water balance.
D. General overview of the relief
For our relief study we have set some Digital Elevation Model (DEM) using computer assisted methods. To each bar of the grille (pixel) where the position is defined by the cartesian coordinates X and Y, we have assigned a numeric value, altitude Z, given by a real number. 
We have established some DEM with different resolutions; for instance figure 4 enables us to have a comprehensive view of the region under study. We can easily observe the Ait Abdi plateau ranging from 2200 to 3000 meters as well as the more uneven one located southward the Taghia village. The large bars of limestone massifs, which form Jbel Aroudane (3359 m) and Jbel Timghazine (3328) are equally well visible. To the north, we find the barrier formed by Jbel Laqroun (3117 m). The deep Assif Ahansal valley, crossed by Toarcian Aalenian marly limestones, and the canyons cutting the limestones to the south of Taghia and to the east of the plateau, well evidence the young age and the dynamism of the relief in this region. 
I.	Exokarst morphology and its evolution
Lithologic and structural conditions, associated with past and present climatic conditions on the Ait Abdi plateau, determine a typical karst morphology which can be described under different observation scales.
A. Dry valleys and "wave karst"
× 	Dry valleys
At the plateau scale, the disorganisation of the hydrographic network and its endorheic character which characterise the karst model clearly appear on the enclosed geomorphological maps. One can find, however, some "organised" dry valleys. The valley bordering the plateau to the West are less hierarchized and shorter than those, facing East, which join Assif-n-Tafraout. This asymmetry seems to be essentially due to the plateau structure with the syncline axis slightly dipping eastwards, and to a facies change in the same direction: in this way the limestones become more closely stratified and marl rich, and accordingly less favourable to karst morphogenesis. (Couvreur, 1988, p.155).
Once the geomorphological map of the plateau has been established, we have digitalized the dry valley network, which we have statistically analysed as we have done with the fracture system. Here we will present the results of this analysis. The general orientation of these organised valleys is also conditioned by the syncline structure: NWward of the plateau, the valleys have an orientation N300-N315 while NEward they have an orientation N30. These two directions correspond to two families of fractures observed on the plateau.
As already noted by Couvreur (1974), the bottom of these valleys is often constituted, especially upstream, by a series of landings, which consist of elongated dolines with a flattened turfed bottom tipping over one from the other (Fig.5). Outflow traces are rare. More one comes closer to the cliff, the more the valleys seem to be sinking down. Both sides of the talwegs take a V shape and the turfed dolines give place to a dry bed clattered with blocks. Finally, to the West the dolines are abruptly cut by the cliff, while in the eastern part of the plateau such tiers are less frequent, and, in case of flooding, temporary outflows are more relevant and active in their erosion process. The cause for this is to be found in the structure and in the lithology of the plateau under study.
At the centre of the plateau, the valley disappear being replaced by doline chaplets resulting from the disorganisation which is due to karstification of an ancient river network in connection with the fracture system.
× 	"Wave depression" karst
In the central part of the plateau, there is a typical karst morphology, marked by a very interesting presence of nival karst, described by Couvreur (1974), and known as "wave depression" karst. This particular landform shape consists, on the slopes, of a series of asymmetric and parallel small valleys1 presenting no discharge and, on the flat bottom, it consists of a regular series of asymmetric dolines with approximately 30 m diameter. Both valleys and dolines have a steep 5 to 10 m high side facing E-NE, and a slightly leaning side facing W-SW. This asymmetry is determined by a series of several structural, lithologic and topoclimatic factors. The upper part of the Ait Abdi formation is constituted by a limestone complex bended to form a slight syncline whose axis dips westwards. The majority of "wave depressions" are concentrated in this western part, their average spacing ranging from 50 to 100 m. They become less and less frequent in approaching the eastern canyon. In that sector, the slopes turn to NE-E and the mini-"cuesta" aspect of those landforms cannot express itself in the same way as in the western part. Still the forms remain as well as their asymmetry (Fig.6). Therefore, structure is not the only cause for that particular morphology. The asymmetry of valleys and dolines can be explained partly by morphodynamic mechanisms affecting the leaning limestone layers (Aubert, 1969, p.334), and partly by exposure conditions (Couvreur, 1988). The snow covering the plateau during the winter time seats in different ways: the slopes exposed to the wind2 are quickly cleared, and therefore they are subject to the night frost which regularise them; the more protected sides (lee side) keep the snow-drifts accumulating there which defend them from the frost-bite3 for a long time (Fig.7). These, however, are eaten up by the more aggressive snow-melting derived waters which penetrate into the steep slopes mostly covered by karren. Maire (1990) made similar observations for Crete, Taurus and Zagros, where he pointed out the role of snow-drifts in the depression asymmetry, as well as Sauro (1973, 1974) in the Lessini mountains in Italy. The structure role appears necessarily unclear: if it is the snow-drifts, caused by the prevailing winds, to be held responsible for the "wave depression" karst landforms, these forms present a profile which is deeply affected by the layer steepness. Figure 6 clearly shows the structure role observed in the field . The dominant processes are well outlined on each of the slope sides by examining the microforms; the steep lee side slopes present several dissolution forms (Rillenkarren, Wandkarren...) while the less steep slopes are littered with frost shattering.
With regard to the general orientation of the plateau blind valleys (which generally end up in closed depressions) one can note that the main valleys, where there is an abundance of wave depression, present the two main orientations observed for the "organised" valleys whereas the orientation of the "wave depression" is submeridian for the reason already explained. In order to confirm this first visual impression, we have made a statistic analysis of the "wave depression" karst directions which we have subsequently digitalized on the geomorphological map.
Figure 5 shows another aspect of the slope asymmetry at a minor scale than the "wave depression" one. The southern slope is better regularized than the slope with small valleys and "wave depressions" which is turned northwards. Couvreur (1974 and 1988) and Maire (1990) talk of a "adret/ubac" climatic effect. If, according to the example under consideration, steepness appears to be an important factor, we should point out that, on the plateau, we can find some regularized slopes in very different structural contexts as, for instance, on the northern side of the great depression located at the Jbel A'Labli foothills. This supports the explanation proposed by the above mentioned authors. The slopes with an exposure to the south are more subject to experience sharp thermal excursion between day and night. The frost shattering is more severe since snow quickly melts away. On the other hand, the slopes facing north are subject to corrosion due to a larger amount of snow-melting derived waters which are more aggressive (Maire 1990).
× 	Statistic analysis of "wave depressions" karst.
The "wave depression" directions on the plateau can provide us with information regarding lithological changes and morphogenesis. We can obtain such information in two different ways:
¥	topographic map analysis: unfortunately the plateau is not covered with a cartographic survey which makes use of the same scale. The western part has a 1:50,000 map while for the eastern part we have only 1:100,000 maps. For this reason, because of the small scale map generalisations, data are not homogeneous for the whole plateau. Accordingly, we have decided not to use such technique for our research study on dry valleys and "wave depression karst" directions.
¥	photogrammetry: a computer-assisted photogrammetry has allowed us to digitalize the plateau "wave depression" network - starting from the scanned air photographs which were subsequently geometrically corrected - directly on the computer screen as soon as we had established the geomorphological map. This technique has the advantage of providing us with a geometrically correct topographic representation. On the other hand, after some manipulation, the digitalized vectors can be used for a statistic program.
The direction and length analysis of these geomorphological components has given the following results (Fig. 8):
We have represented the rose-window of the "wave depression" karst direction frequencies. What we have noted in the previous paragraph is clearly emphasised on this graphic: the distribution of the "wave depression" directions is unimodal, that is all the wave depressions present the same average submeridian direction. This distribution is not in accordance with that of the dips and it demonstrates the weak bond existing between the direction of the" wave depressions" and the geological structure. On the other hand, the direction of the western sector dominant winds can be compared with our rose-window. 
B. The other karst landforms
× 	 Dolines
The Ait Abdi karst plateau presents several close depressions, which we can distinguish in the following types:
¥ asymmetric dolines on an even bottom tied to the "wave depressions" (on which we have dealt so far). They generally present an oval shape, their large diameter being parallel to the "wave depressions";
¥ dolines and ouvalas in dry valleys which we have already mentioned (see fig.5);
¥ dolines sitting on anticline axes and passes, which are affected by extension-decompression fractures as those described by Maire (1990) for Djurdjura in Algeria and for Mt. Ida in Crete.
Almost the totality of these depressions have a bucket form with the exception of pass dolines and the anticlines whose form is funnel-like. On the 186 km2 plateau surface the density is 4.73 dolines per km2 which is relatively low if compared, for instance, with tropical karst. However, on the geomorphological map it appears that the density is not uniform and that some central sections of the plateau are literally dotted with dolines. These are precisely the sectors where the "wave depressions" are highly represented.
× 	Large depressions
Large depressions with flat bottom are distributed on the whole plateau surface. Their location is mainly tied to structural conditions (breaks, faults, overthrusts, dykes). All this can be perfectly observed on the geomorphological map. The majority of this kind of poljes is tied to tectonic undulations which either cross or border them. These are the dykes which are responsible for the largest and most typical landforms such as for instance Almou-n-Ouhanad4. All this results in the elongated shape of these large depressions. 
Some of these "poljes" can contain a small temporary lake during the thaw season. We have observed this phenomenon in the Almou-n-Ouhanad polje in spring where a spring supplies a small brook which in turn fills up a small lake located a few hundred meters apart. We have not observed ponor presence in that polje, the self-emptying lake because of evaporation and diffuse infiltration in the sediments filling the polje. On the other hand we have discovered a typical ponor not far from Almou-n-Ouhanad where its functioning, inhibited by the climate, can be traced back by means of some outflow traces.
The fillings of these depressions can contain a small suspended aquifer. In this same depression a well has been excavated receiving the underground water. It appears that this aquifer is directly linked with the karst aquifer since some shepherds have told us that the wells which outfill in winter create small streams of water supplying the lake at the bottom of the polje. In particular hydrologic circumstances (high waters) this "polje" could occasionally function as level evidence of the karst aquifer. The effective altitude of the bottom of the different plateau polje could correspond to a paleo-piezometric surface belonging to a period more humid than the Quaternary such as that which today regulates the height of the Slovenian poljes (see the base level poljes described by Ford & Williams, 1989). This ancient piezometric surface could have been conditioned by the karst level of the period in which the plateau was not yet a close isolated by the water system compartment as it is today. We believe that this ancient ground level correspond to the Antre des Craves, ancient forced conduit of large dimensions cut by the cliff located on the NW part of the plateau.
It follows that today, the majority of these landforms are "dead poljes", if we accept Couvreur's definition (1988, p.151), which means that their hydrologic functioning is no more active or it is visibly slowed down. Evidences of this hydrologic functioning are still visible: ponors, pits (with or without speleothems), etc...
× 	Karren  
On a large scale, it is equally possible to observe all sorts of karren of high alpine karsts such as those described for the mountains located in temperate latitudes (Boegli 1960; Nicod 1972, Ford & Williams 1989; Maire 1990) and for the  North African mountains. (Quinif 1976, 1977). The limestone lithology can affect the purity of landforms. Marly limestones develop karst landforms (especially microforms) which are less clear-cut, less developed and more subject to frost shattering.
All the microforms of the Ait Abdi plateau which have been studied and catalogued so far correspond to a high mountain karst whose evolution speed is slowed down by the climate aridity.
C. Karst genesis and morphogenesis
The morphogenetic history of our plateau seems to correspond to the one described by Maire (1990, p. 587) for the Mediterranean karst: "We are dealing with supraforest or discontinuously forested karst, with doline fields and ancient talweg networks transformed into depression chaplets (doline valleys) which do not necessarily show clear evidences of the ancient impermeable covering. They were not subject to glacial erosion." Then, the karstification of the Ait Abdi plateau began after the disappearance by erosion of its non karstifiable geologic covering. On the Zawyat Ahansal geologic map, Jossen (1990) has established a palinspastic profile referring to the Bathonian which, at the top of the Ait Abdi limestone formation, shows us the presence of a thin sedimentary continental pellicle of sandstone, conglomerates and red silts, the Gettioua sandstone formation of the Bathonian age (or of a later period if we give credit to Sigogneau-Russel et al. 1990). This seems the only detritic sediment layer which has covered our karst plateau (even if there is disagreement on this issue). On the same map, we find no evidences5 of the "calm" Aptian transgression (Monbaron 1981, 1982) which followed the emerging and shaping phases of an erosion surface (upper Jurassic?-lower Cretaceous) or later formations (Monbaron 1988, Monbaron et al. 1990). Could they have been entirely eroded or not even deposited on this Atlas region? A research focusing more specifically on endo and exokarst fillings will certainly clarify these issues.
After the compression phase, the uplift of the Atlas during the Miocene determined an intense mechanic erosion of the massifs so that the debris was deposited on the foothills (Dir) and in small subsidence basins located inside the mountain range (Cathedral conglomerates, Monbaron 1985, Jossen 1990). During the Mio-Pliocene the non karstfiable covering of the Ait Abdi plateau had probably already disappeared since the majority of pebbles forming the conglomerates is made of limestones of the Jurassic age. More humid and warmer conditions than those present today have carried along the digging of the main valleys to the point that the hydrographic network on the plateau has been "reorganised" by reaching the karstifiable rock level.
With regard to the genesis of the Ait Abdi plateau landforms, Couvreur (1988) distinguishes three phases not datable with precision: 
¥	A phase of reorganisation6 of the hydrographic network,  due to the disappearance of the covering of non karstifiable rocks, appeared after the emerging of the massif which "isolated" the karst by means of deeper and deeper valleys (infra).
¥	The following phase, which certainly came after a more humid period, was characterised by the formation of large poljes located on or close to tectonic undulations, breaks, faults, and overlaps. At the time there was certainly digging and an intense cavities speleothems deposition which today are cut again by the topography. We believe that this phase might correspond to the speleogenesis of the Antre des Craves phreatic gallery (cf pt III).
¥	The last phase, which is close to the present, and it is still ongoing, is marked by a slowing down of the morphogenetic processes due to a progressive aridity of the climate.
It is during this last phase, which we collocate in the Recent Quaternary, that the plateau meso and microforms (dolines, karren) have been created, that some slopes have been regularized while other have been organised in "wave depressions" (table 1).
Couvreur (1976) estimates that the initial digging of the main valleys, up to a range  of 600 m, should be located between the Mio-Pliocene (deposits of the Cathedral conglomerates) and the Plio-Villafranchian. Dating is based on the altitude position of the different plio-villafranchian terraces of the sector relative to the Mio-Pliocene deposits. A further post-villafranchian deepening was probably still active  up to the mid-Quaternary. It was this deepening, which Couvreur estimates in the range of one hundred meters, which was responsible for the canyon digging of our sector.
II. THE ENDOKARST
A speleologic prospect has been carried out by Jacques Brasey7 in the NW plateau region on the base of information provided by the local inhabitants, the Berbers. All the cavities more than 5 m wide have been catalogued and topographied, and their position was reported on the geomorphological map. It should be noted that none of the vertical cavities has reached the water table.
A. Types of cavities
The majority of the cavities discovered are relatively short and they are located on subvertical fractures. However, a few 30 to 40 m deep shafts provide access to rooms of important dimensions, richly decorated, and always obstructed by debris or chemical fillings. As a matter of fact an important speleothems deposition phase, no more active under present conditions, can be found in most plateau cavities and it is frequently visible on the surface because of karst erosion. On the other hand, several cavity entrances are constituted by ancient shafts cut by the present topographic surface. This is the case of the 68 m deep hole, the deepest on the plateau, (AA15, figure 9, situation on the enclosed geomorphological map) whose edge present important speleothems. Other shafts, which are not so deep, are very large as for instance AA4, and they provide access to large rooms draped with ancient speleothems.
On the cliffs bordering the plateau, several holes are clearly visible but they are hardly accessible. The most important of them, Antre des Craves, has an huge 30 per 10 m entry. The gallery, which follows it, is an ancient phreatic gallery obstructed by 2 m thick fillings sealed by some speleothems (figure 10). Unfortunately, after approximately 150 m, the sediments entirely fill the gallery. No outflow is presently visible. This is the gallery we mentioned when we pointed out the ancient level of karst origin which allowed the development  of "base level poljes". Three important clay fillings interspersed with speleothems support this hypothesis.
40 m further down, at the cliff foothills, and 500 m to the north, we find Aghbalou-n-Wattouf spring originating from a 5-8 m large phreatic gallery with a vadose entrenchment (see AA17, figure 11). It is possible to go upstream for about 200 m before the river disappears into a siphon. A second floor (not topographied) is laid upon the river but it does not allow to circle around the siphon. The cavity is developed on the impermeable level of Toarcian Aalenian marls. Besides these explorable cavities, on the plateau we find a large amount of ancient speleothems cut by the present topographic surface. We have also found the complete shape of a circular shaft completely filled and cut by the topographic surface. Maire states (1990, p. 447) that this kind of speleothems are often found in Mediterranean karsts (Crete, Spain etc..); they represent an ancient karstification, favoured by a warmer climate with biopedologic covering, and a significant removal of the underlying layers.
B. Localisation of the cavities and relations with the structure
All the cavities discovered on the plateau belong to the vertical transfer layer of the karst system. The subtabular structure, the important thickness of the formation, and the absence of impermeable levels in the latter have all favoured the development of several vertical and parallel shafts instead of favouring an interconnected karst network. Thus it results difficult to determine, on the base of these observations, the horizontal directions of the outflow since we have no access to the gallery area. The only usable data come from the uncovered cliff cavities. Aghbalou-n-Wattouf cave is developed on two families of subvertical breaks with orientation N160-167 and N016-025. The shafts observed on the cliff are developed approximately on the same families of subvertical breaks. Since the gradient is not relevant (generally < 10¡) it is the potential conditions that affect the water outflow in these families of fissures. Unfortunately we do not have any information on the latter (drilling, wells, tracing tests...).
C. Speleogenesis
The presence of shafts, at times cut by the topographic surface, and of rooms filled with rich speleothems, demonstrates the past existence of a thick soil and a much less pronounced aridity. In fact, it has been demonstrated that the relevance of the corrosion process is proportional to water volumes passing through the karst, that is to the effective infiltration as well as CO2 dissolved into the waters (role of pedologic CO2).(Ford and Williams 1989, p102; Maire 1990,p.415). Cavities of such relevance cannot be created any more by the climate presently observable in our sector of study since it is too dry. Furthermore, Maire (ibidem), Ek (1966) and Meneghel & Sauro (1990) show clearly the bond existing between the average altitude of the catchment area and the water mineralization. In fact, one can observe a reduction of the latter in presence of altitude in relation with a pCO2 reduction (forest and soil disappearance). Our sector of study, as we have already seen, has been subjected to the Atlas orogenesis which has pushed it up. Before then, the climate should have been warmer and the altitude less relevant. A cryptokarstification could be produced when the covering of detrital rocks was still present, quickly followed by a "normal" karstification once the limestones outcropped.
The morphogenetic analysis of Aghbalou-n-Watouf cave and Antre des Craves show at least three distinct phases of karstification:
¥	a first very ancient phase corresponding to the Antre des Craves gallery, about 100 m above the present day outflow level, and to the important phase of speleothems deposition found on the plateau;
¥	a second intermediate phase, materialised by the Aghbalou-n-Wattouf upper gallery, about 20 m above the present underground river;
¥	a third recent phase responsible for the gallery evolution in correspondence with the present day underground river drowning.
It should be noted that, besides the presence of karst drains at that level, different morphologic and lithologic observations in the cliff located to the NW of the plateau lead us to think that the base level of the first phase should have corresponded to a marly level located on the mid wall. In fact, at that level we can observe the following factors:
¥	the frequent existence of a level used by the shepherds to gather up the rams at night (paths arranged on break areas lead there). This form is the product of the joined action of cryoclasty and erosion, which along with gravity, contribute to carve the wall in favour of marly passes (Maire 1990, p.352);
¥	the presence, due to a local marly aquiclude, of a series of small and often perennial springs, or of active oozes during snow melting. These outflows favour the phenomena described above;
¥	the notches cut by the cliff dry valleys descend to the same marly level, independently from the altitude. The digging of these valleys could have been produced, starting from that base level, by regressive erosion, during a more humid phase.
D. The endokarst fillings
Our late lamented colleague Jacques Brasey made several cuts in detrital deposits which he prospected and sampled in different cavities. Sample drawing of speleothems (stalagmites, stalactites) were carried out as well. After his accidental death we have taken some of his observations which we like to present in the following paragraphs.
×	Speleothems
Dating of speleothemes, drawn in June 91, was made by Jacques Brasey with Dr. P.L. Smart at Bristol in January 1992. Ten samples were prepared and dated. The results (table 2) show very different ages but they confirm some morphologic observations:
¥	some samples have an indefinite age, in spite of the method limit of 400,000 years8 (phase 1).  This attests the existence of a middle Pleistocene speleothems layer as well as of a more ancient karstification phase (Tertiary or Plio-Villafranchian).
The other dates show different speleothems deposition phases:
¥	an ancient phase 2 dating 210,000-220,000 years corresponding to a antetensiftian interpluvial (Moroccan continental quaternary chronology, Choubert & al. 1956, Beaudet & al. 1967) in relation with the interglacial which preceded the recent European Riss (see Maire 1990, p.471, figure 276 and p. 481 table 54).
¥	a phase 3 dating 140,000 years corresponding to the pluvial Tensiftian (recent alpine Riss);
¥	more recent phases 4 and 5, dating 50,000 and 30,000 years respectively, corresponding to the pluvial Soltanian (=WŸrm). These different speleothems deposition phases necessarily imply the development of different karstification phases.
¥	a phase 6 of recent Holocene speleothems deposition (3,000 and 5,000 years) was determined by means of a stalagmite drawn from some blocks collapsed in the back of a small room (AA8). This important sinter layer is present in the majority of explored cavities where it often covers the debris and the fallen blocks.
The first sample (AA1-10.9.90-I) comes from a speleothem drawn from the surface. As we have already said, such speleothems deposition layers are often very ancient, and the age determined in this way could be questioned. Maire (1990), using the same method, dated one of these speleothems levels coming from the central calcareous High Atlas which was drawn from Jbel Ghat, located 80 km to SW  of our sector. Curiously that sample provided us with a dubious age (about 140 ka) due to a detrital pollution (relation 230Th/232Th = 5, too weak) which altered the dating. In the case of our sample the relation (230Th/232Th = 35.1 ±6.24) is correct. Therefore we can regard the result obtained as plausible. 
The stages of the network sinking seem to match with the results of the dating. Therefore, there will be three gallery levels, corresponding to three  sinking stages, and a minimum of six speleothems deposition phases. Supplementary dating and a correlation with the detrital fillings remain to be done to confirm these first hypotheses.
If one compares the dating with the Quaternary chronology proposed by Maire (1990, p.471) for the Mediterranean karst, it should be noted that some of these ages correspond to periods where glacial conditions (Riss or WŸrm) dominated the humid temperate domain, conditions which prevented or slowed down the speleothems deposition. In the South Mediterranean region, the Quaternary glaciations did not upset the climatic conditions so radically as on the Alps. Practically there were no glaciers in Morocco (there are only few small glacial cirques visible on the Mgoun north face (Couvreur 1978). The conditions cooled down especially after the Tensiftian (Couvreur 1988), but not enough to stop the speleothems deposition in the caves.
III. CONCLUSIONS
In this article we wanted to sketch an outline of the present relief situation in the sector under study. Computer assisted techniques have certainly helped us in our task. We have seen that there is no direct relationship between the "wave depression" direction and the geological structure, the latter being of relevance only in the slope shape of the dry valleys. We have pointed out the climatic influence  (dominant winds, snow, exposure) in the particular morphogenesis. The nature of the nival karst of the plateau well evidences the variety of the karren forms. Tectonics and hydrogeology have been important factors in determining the genesis of the plateau poljes which today, with few exceptions, do not deserve this designation any more. We have attempted, on the base of purely morphologic criteria, to define the morphogenetic phases of our sector which started during the Miocene, acquired relevance after the maximum raising of the chain and continued in spite of the climatic fluctuations of the Quaternary. Today's phenomena of karst morphogenesis are slowed down by relative climatic aridity. Endokarst studies carried out, and unfortunately left incompleted by Jacques Brasey,  have provided us with precious information on the chronology of the plateau karst evolution; we can distinguish at least three karstification phases and six phases of speleothems deposition. All these data should be furtherly completed to reach a better understanding of the evolution of this karst during the Quaternary.
References
AUBERT, D. (1969), PhŽnom�nes et formes du Karst jurassien, Eclogae geol. Helv., Vol. 62/2, p. 325-399.
BEAUDET, G., MAURER, G., RUELLAN, A. (1967), Le Quaternaire marocain. Observations et hypoth�ses nouvelles, Revue de GŽographie Physique et de GŽologie Dynamique (2), Vol. IX, Fasc. 4, pp. 269-310, Paris.
BOEGLI, A. (1960), Kalklšsung und Karrenbildung, Z. Geomorphol., suppl. Band 2, p. 4-21.
BRASEY, J., PERRITAZ, L. (1992), Recherches gŽomorphologiques et spŽlŽologiques sur le plateau des A•t Abdi (Haut Atlas central, Maroc), Actes du 9�me congr�s spŽlŽologique national, Charmey, SSS.
CHOUBERT, G., JOLY, F., GIGOUT, M., MAR‚AIS, J., MARGAT, J. et RAYNAL, R. (1956), Essai de classification du Quaternaire continental du Maroc, C.R. Acad. Sci., Paris, t. 243, n¼ 5, p.504-506.
COUVREUR, G. (1974), Le r™le de la neige dans l'Žvolution des formes karstiques de haute montagne du Haut Atlas central (Maroc) - MŽmoires et Documents du CNRS, Nouvelle sŽrie, vol. 15, PhŽnom�nes karstiques, tome II, 135 - 138.
COUVREUR, G. (1976), Le creusement des vallŽes dans le Haut Atlas central calcaire (Maroc), Actes du symposium sur les versants en pays mŽditerranŽens, Aix-en-Provence, 1975, CEGERM, vol.V, 1976, p. 109-111.
COUVREUR, G. (1978), Les limites de l'action du froid, actuelles et quaternaires, dans le Haut Atlas central (Maroc), Actes  du colloque sur le pŽriglaciaire d'altitude du domaine MŽditerranŽen et abords, Strasbourg - UniversitŽ Louis Pasteur, 12-14 mai 1977, Association gŽographique d'Alsace, pp. 115-121.
COUVREUR, G. (1988), Essai sur l'Žvolution morphologique du Haut Atlas central calcaire (Maroc) - Notes et MŽmoires du Service GŽologique n¡ 318, Rabat, 391 p.
EK C. (1966), Faible agressivitŽ des eaux de fonte des glaciers: l'example de la Marmolada (Dolomiti). Ann. Soc. Geol. Belgique, t. 89, 177-188. 
FORD, D., WILLIAMS, P.(1989), Karst geomorphology and hydrology, Unwin Hyman, London, 601p.
JOSSEN, J.-A. (1990), Carte gŽologique du Maroc au 1:100'000, Feuille Zawyat Ahan�al, Notes et MŽmoires n¼ 355, Rabat.
MAIRE, R. (1990), La haute montagne calcaire, Karstologia-MŽmoires n¼3, AFK, FFS. 731 p.
MENEGHEL M. & SAURO U. (1990), Reflection about high mountain karst environment and their fragility in the Dolomites. Acta Carsologica, XIX (1990), 207-214.
MONBARON, M. (1980), Le magmatisme basique de la rŽgion de Tagalft, dans son contexte gŽologique rŽgional (Haut Atlas central, Maroc), C. R. Acad. Sci. (Paris), (D), 290,  pp.1337-1340.
MONBARON, M. (1981), SŽdimentation, tectonique sysnsŽdimentaire et magmatisme basique: l'Žvolution palŽogŽographique et structurale de l'Atlas de Beni Mellal (Maroc) au cours du MŽsozo•que, ses incidences sur la tectonique tertiaire, Eclogae geol. Helv., vol.74/3, pp. 625-638.
MONBARON, M. (1982), PrŽcisions sur la chronologie de la tectogen�se atlasique : exemple du domaine atlasique mŽsogŽen du Maroc, C.R. Acad.Sc.Paris, t. 294, 19 avril 1982.
MONBARON, M. (1985), Carte gŽologique du Maroc au 1:100'000, Feuille Beni Mellal, Notes et MŽmoires n¼ 341, Rabat.
MONBARON, M. (1988), Un serpent de mer: le probl�me de la datation des "Couches rouges" du Haut-Atlas marocain. Le point de la situation, Actes de la sociŽtŽ jurassienne d'Žmulation, 1988.
MONBARON, M., KUBLER, B., ZWEIDLER, D. (1990), Detrital rubified sedimentation in the High Atlas Trough in the Mesozoic: attempt at lateral correlation using correspondence factor analysis , Journal of African Earth Sciences, Vol. 10, No 1/2, pp. 369-384, 1990.
MONBARON, M., PERRITAZ, L. (1993), Le  plateau des A•t Abdi (Haut Atlas central, Maroc), un mod�le de karst perchŽ d'altitude en zone semi-aride,Third International Geomorphological Conference, Hamilton, Canada, unpublished.
MONBARON, M., PERRITAZ, L. (1994), Ressources en eau d'un karst marocain, Universitas Friburgensis, n¼ 2, Fribourg, pp.19-22.
NICOD, J. (1972), Pays et paysages du calcaire, PUF, Paris.
PERRITAZ, L. (1995), Contribution ˆ l'Žtude gŽomorphologique et hydrogŽologique d'un karst perchŽ en domaine mŽditerranŽen: le plateau des A•t Abdi (Haut-Atlas central Maroc), PhD thesis, University of Fribourg, Switzerland, 178 p.
PERRITAZ, L. (1996), Approche gŽomorphologique du plateau karstique des A•t Abdi
(Haut Atlas central, Maroc), Karstologia no 28, AFK, SFS, Bordeaux.
QUINIF, Y. (1976), Contribution ˆ l'Žtude morphologique des karsts algŽriens de type haut-alpin, Revue de GŽographie Physique et de GŽologie Dynamique (2), Vol. XVIII, Fasc. 1, Paris, pp. 5-18.
QUINIF, Y. (1990), La datation des spŽlŽoth�mes (U/Th) appliquŽe aux sŽquences sŽdimentaires souterraines pour une mise en Žvidence des ruptures palŽoclimatiques, Karstologia MŽmoires n¼2, Actes du colloque "Remplissages karstiques et palŽoclimats", Fribourg, 13 et 14 octobre 1989, FFS, AFK, p. 23-32.
SAURO U. (1973), Il Paesaggio degli alti Lessini. Studio geomorfologico. Museo Civ. di St. Nat. di Verona, Mem. f. s. , 6, 161 pp. 
SAURO U. (1974), Casi di dissimmetria climatica in vallette degli Alti Lessini. Natura e Montagna, 1974, 2, 29-32. 
SIGOGNEAU-RUSSEL, D., MONBARON, M., DE KAENEL, E. (1990), Nouvelles donnŽes sur le gisement ˆ mammif�res mŽsozo•ques du Haut-Atlas marocain, Geobios, No 23, fasc.4, pp. 461-483.