Based on data presented in this study, the E-W–trending Alima anticline in the Metlaoui region of the southern Tunisian Atlas Mountains formed due to far-foreland, brittle deformation. The Alima anticline is one in a series of en echelon folds in the Atlas fold-and-thrust belt of North Africa. Geologic mapping indicates that the Alima anticline has a steep southern limb, a gently dipping northern limb, and pervasive normal fault sets. Fracture orientations suggest that fracturing occurred early in the fold history as a synfolding process, not as a pre- or postdeformational process. Gravity data show positive Bouguer anomalies near fold crests, not the negative anomalies that would be expected if the anticline were salt cored. Seismic data, collected along lines in basins surrounding the Alima anticline, suggest the presence of several high-angle reverse faults. Based on surface and subsurface studies, we attribute the development of the Alima anticline to far-foreland deformation associated with late Cenozoic contraction. N-S–directed elongation in the Triassic reoriented to NW-SE–directed shortening in the Miocene, causing Triassic normal faults to be reactivated as oblique-slip reverse faults. A comparison of the Alima anticline to other anticlines in the region suggests that several different styles of folding are present, each representing a different time of initiation.

In the far-foreland of mountain belts, including the Sunda Shelf of Indonesia, the Alpine Mountains of Europe, the Algerian Atlas Mountains, the southern North Sea, and the Rocky Mountains of the western United States, fault reactivation is documented as a mechanism for fold development (e.g., Ziegler, 1983; Letouzey et al., 1990; Wise and Obi, 1992; Vially et al., 1994; Remmelts, 1995). Preexisting faults have long been acknowledged as zones of weakness in crustal rocks during subsequent tectonic activity, and they are commonly favored for reactivation over new fault generation (e.g., Donath and Cranwell, 1981; Etheridge, 1986; Holdsworth et al., 1997). Factors such as the orientation of the fault, the coefficient of sliding friction, and the state of stress are thought to control the conditions under which a fault is reactivated (Letouzey, 1990). Examples of fault reactivation have been documented in physical experiments (e.g., Koopman et al., 1987; Buchanan and McClay, 1991; Sassi et al., 1993), in seismic data (Ziegler, 1983; Etheridge et al., 1985; Letouzey et al., 1990), and in the field (e.g., Holdsworth et al., 1997).

In order to study reactivated faults directly, they must breach the surface and retain kinematic indicators from multiple episodes of activity or retain changes in deformation products such as cataclasites within breccias (White, 1986; Holdsworth et al., 1997). The presence of salt can further complicate the causal relationship between reactivated faults and folding. The movement of salt through the subsurface can generate salt-cored anticlines, which can be difficult to distinguish from folds formed from fault reactivation. Thus, it is necessary to gather evidence from stratigraphy (e.g., Etheridge, 1986; White, 1986), geometric similarity (e.g., Etheridge, 1986; Holdsworth et al., 1997), and geophysical data to deduce whether fault reactivation is a viable mechanism for fold formation. Additionally, folds associated with fault reactivation may exhibit synchronous, local brittle extensional structures and particular joint patterns (Unruh and Twiss, 1998; Cooper et al., 2006).

This study examines the E-W–trending Alima anticline to investigate folding mechanisms in the Metlaoui region of the southern Tunisian Atlas Mountains. The Alima anticline is one in a series of en echelon folds in the distal Atlas fold-and-thrust belt of central Tunisia. Geologic mapping, fracture data, gravity data, and seismic data are utilized to establish whether far-foreland, brittle deformation associated with the Alpine orogeny generated folds in the Metlaoui region. In addition, these data are used to investigate whether salt diapirism or fault reactivation played a larger role in the generation of the Alima anticline and similar folds in the southern Tunisian Atlas Mountains. Geophysical data are then used to establish episodes of folding since the Cretaceous.

The Atlas Mountains extend 2000 km from the Atlantic margin of Morocco to the Mediterranean coast of northern Tunisia (Fig. 1). The Tunisian Atlas Mountains are bounded to the north by the Tell-Rif chain, to the south by the Saharan Platform, and to the east by the Sahel block (Fig. 1; Ben Ferjani et al., 1990; Burollet, 1991; Swezey, 1996; Hlaiem, 1999; Bouaziz et al., 2002).

In Tunisia, the Atlas Mountains began as a series of failed rift basins associated with the breakup of Pangea and opening of the Neotethys from the Permian to the Early Cretaceous (Laville and Petit, 1984; Winterer and Hinz, 1984; Andrieux et al., 1989; Stets, 1992; Grasso et al., 1999; Dhahri and Boukadi, 2010). Although extension occurred throughout this time, two main rifting events are recognized in the Tunisian Atlas Mountains. The first extensional deformation occurred from the Jurassic to the Early Cretaceous and formed E-W–striking normal faults and basins such as the Fault of Chotts and Chotts Basin (Guiraud et al., 1987; Ben Ferjani et al., 1990; Zouari et al., 1999; Soua et al., 2009). The second phase of rifting occurred during the Early Cretaceous and generated NW-SE–striking normal faults and basins, such as the Gafsa fault and basin (Guiraud et al., 1987; Boccaletti et al., 1990; Soua et al., 2009). The development of the basins (grabens and half grabens) resulted in stratigraphic variations in southern and central Tunisia (Bishop, 1988; Bédir, 1995; Hlaiem et al., 1997; Hlaiem, 1999; Soussi and Ben Ismaïl, 2000). In late Early Cretaceous (late Aptian to early Albian), extension was interrupted by a brief period of contraction, termed the “Austrian Phase” (Ben Ayed, 1986; Guiraud et al., 1987; Khomsi et al., 2004), but extension resumed in the Late Cretaceous (Bédir, 1995; Bouaziz et al., 2002).

Although the timing is not well established (Zargouni, 1986; Zouari et al., 1999; Khomsi et al., 2006), Mesozoic extension switched to contraction in the Late Cretaceous–early Paleocene with the closing of the Tethys and convergence of Africa and Eurasia (Bédir, 1995; Gomez et al., 1998; Beauchamp et al., 1999; Hlaiem, 1999; Frizon de Lamotte et al., 2000; Bouaziz et al., 2002). Late Cretaceous to present contraction occurred dominantly in two major events in Tunisia: the Atlasic contractional event and the Alpine contractional event (Tlig et al., 1991; Outtani et al., 1995; Bouaziz et al., 2002; Masrouhi et al., 2007, 2008). The first major event (i.e., the Atlasic event) corresponds to NW-SE to NNW-SSE shortening, and it caused significant inversion of Mesozoic basins during the early and late Eocene (Dercourt et al., 1986; Ricou et al., 1986; Gealey, 1988; Ben Ferjani et al., 1990; Bouaziz et al., 2002; Bracène and Frizon de Lamotte, 2002). The Cenozoic contraction reactivated major Mesozoic normal faults, causing the NW-SE–striking Gafsa fault to reactivate as an oblique right-lateral strike-slip fault (Zargouni and Ruhland, 1981; Zargouni and Trémolières, 1985), and the E-W–striking Fault of Chotts to reactivate as a transpressional feature (Zargouni and Trémolières, 1985; Ben Ayed, 1993; Outtani et al., 1995).

The second major event, the Alpine event, began during the middle to late Miocene and is interpreted to have been the major orogenic episode in Tunisia (Rouvier, 1977; Ben Ferjani et al., 1990; Bobier et al., 1991). During this event, NW-SE–oriented shortening progressively deformed northern and central Tunisia from north to south. Metamorphism and nappe emplacement occurred in the Tell-Rif Chain in northern Tunisia, and basin inversion is interpreted to have formed the vergent NE-SW–trending folds in the northern Tunisian Atlas (Boccaletti et al., 1990). In the southern Tunisian Atlas Mountains, this orogeny again reactivated major Mesozoic normal faults, such as the Gafsa fault (Bobier et al., 1991). Southwest of the Gafsa fault, in the far-foreland of the orogeny, E-W– to ENE-WSW–trending folds developed (Bouaziz et al., 2002; Khomsi et al., 2009). After a period of late Pliocene extension, contraction resumed during the early Pleistocene, reactivating strike-slip faults and Atlas folds (Ben Ayed, 1986; Chihi, 1992; Bouaziz et al., 2002). Minor contractional events continued throughout the Pleistocene, and NW-SE regional shortening continues as a result of slow convergence between North Africa and Eurasia (Dlala and Hfaiedh, 1993; Rebai, 1993; Chihi, 1995; Bouaziz et al., 2002).

This study focuses on the southern Tunisian Atlas Mountains, specifically the Metlaoui region between the Gafsa fault and the Fault of Chotts; the Fault of Chotts is interpreted as a blind thrust fault that cores the Chott Range (Hlaiem, 1999) (Fig. 2). The Gafsa fault bounds the area to the north, and the Fault of Chotts marks the southern boundary. Both faults are major right-lateral oblique-slip features interpreted as reactivated basement faults (Hlaiem, 1999). The Gafsa fault currently displays right-lateral strike-slip movement, although evidence from seismic data suggests multiple past movement directions and reactivations (e.g., Boukadi and Bédir, 1996; Hlaiem 1999). The Fault of Chotts is associated with a number of E-W–oriented anticlines, and evidence from seismic studies suggests several reactivation episodes (Hlaiem, 1999).

The Metlaoui region is characterized by broad synclines and tight anticlines that create large depressions and narrow, continuous E-W–trending ranges, respectively (Fig. 2). The anticlines are south vergent, with southern limbs near vertical and locally overturned, and regional-scale faults are oriented NW-SE and E-W. The en echelon folds that comprise the Metlaoui and Moulares Ranges, including the Alima anticline, are similar in deformational style, with arcuate and segmented, right-stepping fold axes. Fold axes are 8–25 km long, but typically ∼17 km, and anticline widths vary from 5 to 10 km. The anticlines commonly have a second, tighter anticline off the southern flank. The tighter folds are sometimes associated with normal faults, growth faults in Cenozoic strata, and Quaternary alluvium deformed around the anticlines (Ahmadi et al., 2004).

Stratigraphic data (Fig. 3) from the Metlaoui region are derived from well GNT-1 (Fig. 2; Ben Ferjani et al., 1990; Hlaiem, 1999), which was drilled near the center of the Metlaoui Basin, and from stratigraphic data obtained from outcrops around the Alima anticline. Well GNT-1 is 4800 m deep and terminates in Jurassic strata. Combined, the Jurassic and Cretaceous strata are >3900 m thick and are dominated by carbonate and marly shale (Fig. 3). The Jurassic strata consist of relatively thin (350–500 m) limestone of the Nara Formation. The Nara Formation is overlain by Lower Cretaceous carbonate and sandstone (∼1550 m) that are mapped as the Gafsa Group. The Gafsa Group is overlain by the Mansour Group, which is ∼1550 m of Middle Cretaceous marly limestone and shale. The Mansour Group is overlain by Late Cretaceous chalky limestone and shale (550 m), collectively mapped as the Berda Formation. The Berda Formation is subdivided into lower and upper limestone members separated by a middle shale member.

In the center of the Metlaoui Basin, the Mesozoic section is conformably overlain by ∼950 m of Cenozoic strata. In this area, Paleogene lithologies vary from shale and carbonate to evaporites, whereas Neogene strata are primarily clastic. The upper boundary of the Berda Formation represents a major time gap, which is conformably overlain by lower Paleocene black shale that is mapped as the ∼60-m-thick El Haria Formation. Overlying the El Haria Formation, there lie the marl and evaporites of the 80-m-thick Tseldja Formation, which accumulated during the middle Paleocene (Swezey, 2003, 2009). The Paleocene marl and evaporites are overlain by ∼10–20 m of lower Eocene phosphatic shale of the Chouabine Formation, 50 m of chalky/oyster-rich limestone of the Kef ed Dour Formation, and 40–60 m of middle to upper Eocene evaporites and nonmarine shale of the Seugdal Formation. Oligocene strata are not preserved in much of the southern Tunisian Atlas Mountains, and truncated Upper Cretaceous and lower Cenozoic strata are interpreted to be a product of an Oligocene unconformity. Where present, Oligocene strata consist of thin nonmarine siltstone mapped as the Sehib Formation. Miocene strata consist of fluvial sandstone and conglomerate of the Beglia Formation, overlain by upper Miocene to Pleistocene sandstone and mudstone of the Segui Formation, and Quaternary alluvium (Swezey, 1996, 2003).

Location and Exposed Units

The Alima anticline is one in a series of en echelon folds that form the Metlaoui Range, one of the E-W–trending mountain ranges in the southern Tunisian Atlas Mountains (Fig. 2). The Alima anticline is ∼18 km long, 8 km wide, and is located between the towns of Metlaoui to the southeast and Moulares to the northwest (Fig. 2). The oldest stratigraphic unit exposed in the Alima anticline is the Upper Cretaceous lower member of the Berda Formation, and the youngest unit is the Miocene Beglia Formation (Fig. 4).

The crest of the Alima anticline consists of a resistant limestone that is mapped as the upper member of the Berda Formation (Fig. 3). The top ∼30 m of this limestone are exposed in cuts and ravines, and the majority of the structural data in this publication come from this same stratigraphic interval. The upper member of the Berda Formation is eroded by the Tseldja River (i.e., Tseldja Gorge) to the east and in a wide gorge at the western portion of the Alima anticline. In these areas, the lower and middle members of the Berda Formation are exposed. The Kef ed Dour Formation forms a prominent ridge around the exposed anticline margin (Fig. 4).

Primary Fold Geometry

To constrain the primary fold geometry of the Alima anticline, the field area was mapped on 1:150,000-scale topographic maps (Nouvelle Carte de Tunisie, 1993a, 1993b) using the base geology mapped by Regaya et al. (1991) (Fig. 4A); as well, four approximately N-S–trending cross sections were constructed (Fig. 4B). Bedding orientation was measured every 25 m along these sections, elucidating the asymmetry of the Alima anticline (Fig. 4B). The northern limb of the anticline dips 10°–20° northward, and it is cut by a normal fault in the northwestern and north-central portions of the anticline (Fig. 4A). The southern limb dips south, ranging in dip from near horizontal at the fold crest, to 30°–50° progressing southward. Approaching the southern edge, the southern limb is gently folded into a syncline and cut by a normal fault. South of this fault there is a tight anticline (Fig. 4B), the southern limb of which is near vertical or overturned. This feature generally affects the Cenozoic strata but also steps down into the Cretaceous Berda Formation. South of the second, tighter anticline, upper Eocene, Oligocene, and Miocene strata dip steeply south (>60°).

Secondary Fold Geometry

The Alima anticline fold crest is segmented, indicating that it consists of three anticlines. The three subsidiary anticlines are identified by three topographic peaks: Zarrif in the east, Alima in the center, and Zimra in the west. The Zarrif anticline is tightly folded, with a linear fold crest ∼6 km long. In contrast, the fold crests associated with Zimra peak and the Alima peak are more arcuate, trending from E-W to ENE-WSW. The anticlines associated with the Zarrif and Alima peaks are separated by a NE-SW–trending syncline, and the Zimra and Alima anticlines also appear to be offset by normal faulting. The Zimra anticline is a broader anticline with an observed fold hinge approximately ∼7 km long. Overall, the secondary anticlines associated with the Alima anticline show curved, right-stepping, en echelon geometry similar to that observed regionally in the E-W–trending mountain ranges of the southern Tunisian Atlas Mountains (Fig. 2).


Map-scale faults are observed within the Alima anticline (Fig. 4). These faults can be broken into those that occur in the northern/central portions of the anticline, and those that occur in the steeply dipping southern limb. For all observed faults, offset bedding suggests normal motion, although an unobserved strike-slip component may also exist. In the northern/central portions of the Alima anticline, faulting is densest near the secondary Zimra Peak anticline. In this area, faults are generally oriented N-S and E-W and have >10 m throw. The lone exception is an E-W–trending normal fault with >100 m of throw (Fig. 4). Southeast of the Zimra fold crest, faults trend approximately NW-SE, rather than the N-S trend observed in the northwest. Progressing eastward in the central portion of the anticline, normal faults become arcuate, WNW-ESE–trending features. Elsewhere in the northern/central portions of the anticline, minor (meter-scale offset) limb-parallel normal faults are observed.

On the southern flank of the Alima anticline, apparent normal faults occur mostly within the lower Cenozoic strata, but they are occasionally observed in the upper member of the Berda Formation. Apparent offset along these faults in the southern limb varies, but it is generally 30–80 m, as inferred from stratigraphic offset. Whether these faults contain a significant component of strike-slip movement is unknown, but fault steps are commonly joined by strike-slip faults, which are present around the exposed margin of the anticline (Fig. 4). Two dextral faults with relatively large (∼100 m) displacement trend in the same orientations (NW-SE) and are located in the southeastern and southwestern limbs of the Alima anticline (Fig. 4).


Fracture data were collected along the top of the upper member of the Berda Formation. This top surface is an exposed pavement at numerous localities around the Alima anticline. Three stations were established on the northern limb of the anticline, and four stations were established on the southern limb. The number of transects and the length of each transect varied at each station because of variations in local exposure, fracture frequency, and accessibility. All fracture sets are subperpendicular to bedding, and fracture orientations are reported through a series of rose diagrams (Fig. 5). The plotted fracture patterns obtained from the southern flank of the Alima anticline initially appeared random, so the fracture orientations were then rotated. Rotation of the bedding to horizontal was done by a horizontal-axis rotation, where the strike of the bedding is the rotational axis and the magnitude of the rotation is the dip. The average trace lengths for each fracture set at corresponding stations were plotted in bar graphs to illustrate changes in fracture length between stations (Fig. 5). In addition, fracture spacing (the perpendicular distance between a fracture and another fracture within that set) was determined for each transect (Fig. 5).

The results of the fracture analyses indicate that the following two dominant fracture sets are present throughout the anticline: one set oriented approximately N-S, and one set oriented approximately E-W (Fig. 5). Only at station P2 does there appear to be an obvious third fracture set (oriented NE-SW). The fracture sets on the northern limb of the anticline have similar orientations, but the fracture sets on the southern limb show more variation. Zargouni (1986) examined fractures along the southern limb of the Metlaoui Range and noted similar fracture orientations on the southern flank of the Alima anticline. Disregarding station P2, the average trace length of fracture sets tends to increase westward, whereas the average spacing of fracture sets decreases to the west. The longest trace lengths are found at station P2, and the most closely spaced fractures, in addition to the greatest concentration of faulting, are located in the northwestern corner of Zimra peak. Neither of the two fracture sets preferentially transect or terminate into the other (i.e., mutually crosscutting), suggesting the fracture sets formed contemporaneously.

Gravity Data Collection

Gravity measurements were recorded using a Lacoste and Ramberg G-series gravimeter (G19). Associated Global Positioning System (GPS) measurements for each station were recorded using a Leica 9500 GPS receiver. GPS locations for each station were processed using the software program SKI, and base stations were corrected to allow for >0.1 cm (0.04 in.) precision in the horizontal and vertical directions.

Gravity measurements were collected along five N-S–trending transects crossing the Alima anticline and one E-W–trending transect near the eastern Alima anticline fold axis (Fig. 6). Transects ranged from 7 km to 12 km in length, with gravity measurements recorded at ∼400 m intervals along each transect, covering ∼30 km in the E-W direction and ∼15 km in the N-S direction. Raw gravity data were then corrected for tidal effects using the software program QCtool. Base-station gravity measurements were taken at the beginning and end of each collection in order to correct for instrument drift and to tie each daily occupation together. Although kilometer-scale gravity measurements exist for northern Tunisia (e.g., Jallouli et al., 2005) and southern Tunisia in the Telemzan High area (Gabtni et al., 2006), the small-scale and location of our survey did not allow ties into any regional survey. Due to the small scope of the gravity survey, latitudinal gravity corrections were done with respect to the base station. After tidal influences, instrument drift, and latitude were corrected, free-air and Bouguer corrections were applied using a density of 2.67 g/cm3. Terrain corrections were applied using the methods outlined by Hammer (1939) out to the J-ring using unpublished Matlab code (M. Oliver and S. Giorgis, 2006, personal commun.). To correct for regional trends in the gravity data (e.g., Montesinos et al., 2005), each transect's gravity data were plotted along a two-dimensional (2-D) trend. A linear regression was then used to identify the regional trend within each transect. The regional trend was then subtracted from gravity data, allowing smaller anomalies in the data to be observed.

Results of Gravity Measurements

The resulting complete Bouguer gravity anomaly map ranges from −3.24 mGal in the northwest of the gravity survey to +9.50 mGal in the central portion (Fig. 6A). Bouguer anomalies are reported relative to the base station used for instrumental drift corrections. Gravity anomalies shown in Figure 6A are broadly E-W trending, with minor trends in the NE-SW direction. The major feature shown by the gravity map is an elongate, E-W–trending gravity high in the center of the survey. Gravity anomalies decrease on all edges of this feature, except on the NE corner, where a 5 km extension of the gravity high occurs. In addition to the central gravity high, two other gravity highs are observed: one on the eastern margin of the survey area, and the other on the western margin. The gravity anomalies are segmented in the E-W direction, with the E-W–elongate highs separated by narrow, N-S–trending low anomalies. Finally, the major gravity high is clearly delineated from its surroundings, with relative gravity anomalies decreasing in all directions.

Figure 6B shows the relation of the gravity data to the Alima anticline geology. The E-W trend of the gravity data appears to reflect the overall structure of folding in the Metlaoui Range, with its E-W–trending en echelon folds. Although the gravity survey area is largely in the western portion of the Alima anticline, the scale of the prominent gravity high matches the scale of folds in this region. The central gravity high is ∼15 km in the E-W direction and ∼7 km in the N-S direction. Fold axes in this area are ∼17 km long (E-W) and 5–10 km wide. In addition, the segmentation of the gravity data in the E-W direction closely reflects the segmentation of the en echelon folds of this area, with the major gravity anomaly in the central portion of the gravity study occurring in concordance with the Alima anticline. The fold limbs coincide with the gravity high in the western portion of the anticline, indicating that a gravity high is associated with the anticline. However, the major central gravity high does not follow topography, and gravity highs are not observed at the fold crest. The central gravity high occurs in the southwest, and mostly along the southern limb of the Alima anticline. In addition, both the NE-SW–trending gravity high and the gravity high observed in the eastern study area correspond to areas between fold crests, rather than the fold crests themselves.

Construction of Gravity Models

Gravity studies on fold development in the Tunisian Atlas Mountains have focused on the role of salt (e.g., Vila et al., 1996; Jallouli et al., 2005). In order to constrain the involvement of salt in the Alima anticline, 2-D forward models were constructed to fit the gravity data collected in this study (Fig. 7). Forward model results are presented for the following two cases: (1) the Alima anticline without salt (Fig. 7B); and (2) the Alima anticline cored by a salt diapir (Fig. 7B). The 2-D models correspond to the Zimra cross section (Fig. 4) and the N-S gravity transect east of the geologic cross section (Fig. 6). The stratigraphy of the two models is based on well GNT-1 and geologic cross sections of the anticline. Well GNT-1 penetrates Jurassic to Pleistocene strata. However, in the Alima anticline, the Eocene Kef ed Dour Formation is overlain by Quaternary sediment and is thus the youngest stratigraphic unit used in the forward model. Densities were dominantly extrapolated from strata of similar ages to the north (as described by Jallouli et al., 2005), and are shown in Figure 7. Rock descriptions, such as the amount of shale present in limestone, were also utilized in determining densities.

Conclusions from Gravity Models

The model of the Alima anticline without salt was able to reproduce the observed gravity anomaly. The model incorporating salt as a diapiric structure in the core of the Alima anticline produces a negative gravity anomaly. Normal faulting along the fold crest necessarily produces the observed 2 mGal decrease in gravity between the two gravity peaks. However, when salt is incorporated into the model, the low density of a central salt diapir dominates the gravity signature, masking the effects of normal faulting along the fold crest. The comparison of the two models suggests that salt diapirism played a limited role in the development of the Alima anticline.

Benassi et al. (2006) constructed a model in which a salt diapir could produce a relative gravity high similar to our results, but it requires salt to outcrop, piercing unconsolidated sediment at the surface. This geometry is not a viable model for the relative gravity highs in the Alima area since salt does not outcrop. The forward gravity models constructed by Benassi et al. (2006) that include nonoutcropping salt produce a relative gravity low, not the relative highs observed along the Alima anticline. We infer that it is difficult to model the relative gravity highs observed at the Alima anticline if the anticline is cored by salt, in the absence of salt outcrop. The model presented in this study does not preclude the presence of Triassic salt acting as detachment layers within the stratigraphy, but well GNT-1 does not indicate the presence of any salt of Jurassic or younger age.

Seismic Data Collection

Seismic-reflection lines (provided by Compagnie de Phosphate de Gafsa), outcrop data, and published maps provide evidence of growth strata, erosional truncations, and unique folding styles within the southern Tunisian Atlas Mountains. The locations of four seismic transects are displayed in Figure 8, and the processed seismic data and corresponding interpretations are displayed in Figure 9. Seismic transect GSB-150 runs SW-NE, approaches the Gafsa fault, and crosses a portion of the southwestern limb of a Gafsa-related fold at Djebel Bou Ramli. Seismic line Met-8909a runs NNW-SSE and crosses the northern flank of the Chott Range at Djebel Bou Helal. The two remaining seismic lines, Met-8909 and MTO-057, run NNW-SSE and cross two of the less prominent folds in the Metlaoui Basin. Well GNT-1 (Figs. 2 and 3) was used to tie seismic horizons to stratigraphic units. The interpreted seismic lines (Fig. 9) show the following stratigraphic reflectors: the Jurassic–Cretaceous boundary, the Aptian–Albian boundary, the top of the Upper Cretaceous (Turonian) Guettar Member of the Zebagg Formation, the top of the Upper Cretaceous Berda Formation, the top of the Eocene Kef ed Dour Formation, the Eocene–Oligocene boundary, and post-Oligocene strata. The seismic lines have not been converted from two-way traveltime to depth; therefore, dips of stratigraphic units and faults are vertically exaggerated.

The Gafsa fault is currently considered active based on seismic activity and the deformation of Quaternary alluvium (Chihi, 1992; Dlala and Hfaiedh, 1993). Seismic data suggest that folds associated with the Gafsa fault display the greatest amount of deformation. Unlike other folds in the southern Tunisian Atlas Mountains, several faults associated with Gafsa Range folds crop out at the surface, forming dramatic cliff faces and exposing rocks as old as Jurassic and Triassic. It is difficult to visualize the geometry of the Gafsa folds in map view because of the abundance of faulting, but four to five regional-scale folds are associated with the main fault in the western half of the southern Tunisian Atlas Mountains (Fig. 8). The smallest fold has a long axis of ∼8 km and a short axis of ∼3 km. The largest fold has a long axis of ∼21.5 km and a short axis of ∼4 km. In general, the folds are elongate, with long axes parallel or subparallel to the Gafsa fault. The seismic line near the Gafsa fault, GSB-150, shows evidence of several episodes of deformation (Fig. 9A). A steeply dipping fault is just visible on the northeastern side of the seismic line, interpreted to be a part of the Gafsa fault zone. In Figure 9A, the Upper Cretaceous Zebagg Formation thins as it approaches the fault in the NE, and potential tectonic onlaps are present above the Upper Cretaceous (Turonian) Guettar Member of the upper Zebagg Formation. These strata are interpreted to be truncated by an Oligocene unconformity, and there is evidence of growth strata and another unconformity within Miocene strata. Boukadi and Bédir (1996) and Hlaiem (1999) also documented unconformities in Upper Cretaceous and lower Cenozoic strata south of the Gafsa fault, as well as large thickness discrepancies in sedimentary strata across the fault.

Seismic data from the Fault of Chotts also provide evidence for multiple episodes of deformation. The seismic transect Met-8909a crosses the northern flank of Djebel Bou Helal (Fig. 9B). Several onlaps and unconformities are visible in the seismic data. For instance, Aptian strata are truncated by Albian strata, the Upper Cretaceous Berda Formation thins toward the Chott Range, the Upper Cretaceous Berda Formation and the Eocene Kef ed Dour Formation are truncated below the Oligocene unconformity, and internal discordances are present in post-Eocene strata. Hlaiem (1999) also observed unconformities and growth strata in Upper Cretaceous and Cenozoic strata on the northern flank of Djebel Bou Helal (Chott fold), as well as a more deformed southern limb and large thickness discrepancies across the fault in Mesozoic and Cenozoic strata.

Several less conspicuous folds exist in the Gafsa, Metlaoui, and Moulares basins. These folds are generally ∼12 km long and ∼7 km wide, and they are covered by Miocene and younger strata (Figs. 9C and 9D). The seismic lines show the southward vergence of these folds, and the strata below the Eocene Kef ed Dour Formation are parallel with no unconformities, as opposed to strata seen in seismic data from the Gafsa- and Chott-related folds. However, the seismic lines from Met-8909 and MTO-057 are very different from each other in terms of deformational style and growth strata. Seismic line Met-8909 shows a more gentle folding style and lacks growth strata, but Quaternary alluvium is deformed around this anticline (Fig. 9C). In contrast, seismic line MTO-057 shows a steep southern flank and growth strata where Oligocene and younger strata vary in thickness across the anticline (Fig. 9D). Quaternary alluvium is also being actively folded around this anticline. Farther south of the anticline, seismic line MTO-057 shows another structure where a southward-dipping normal fault displaces Lower Cretaceous and Jurassic strata, and Lower Cretaceous strata thicken toward the fault from the south. Above the fault tip, Upper Cretaceous strata are gently folded.

Models for Alima Anticline Development

Several models of folding have been proposed for the southern Tunisia Atlas Mountains. Outtani et al. (1995) and Hlaiem (1999) interpreted anticlines developed in the southern Tunisian Atlas Mountains to result from ramp-related folding that deformed the sedimentary cover over rigid (Paleozoic and older) strata during a single tectonic event. In this context, folding above a Middle Cretaceous décollement led to the development of the Alima anticline (Outtani et al., 1995). However, seismic data refute the fault-bend fold interpretation as it applies to the Metlaoui region. The fault-bend fold model requires the presence of gently dipping thrust faults to form fault-bend folds in the sedimentary cover of the southern Atlas Mountains. Reverse faults associated with the Chott Range have dips >45° (Hlaiem, 1999), not the 10° and 40° expected with fault-bend fold ramps (Serra, 1977; Boyer and Elliot, 1982; Jamison and Pope, 1996). On a larger scale, most folds in the southern Tunisian Atlas Mountains are en echelon and underlain by steeply dipping reverse faults (Chihi, 1992; Bédir, 1995). Furthermore, the faults in the southern Tunisian Atlas Mountains are not confined to the Cretaceous strata. Instead, the faults cut Jurassic strata with no evidence of thrust ramp development in Cretaceous strata.

Detachment faulting along salt beds is an alternate mechanism for fold development in the southern Tunisian Atlas Mountains. Although no salt is observed at the surface in the Metlaoui region, Triassic salt is abundant in northern Tunisia, and a Triassic salt diapir (Hadifa) is present in the Chott Range located east of the Gafsa fault (Snoke et al., 1988; Hlaiem et al., 1997). Accumulations of salt can localize deformation to the thinner overburden overlying the salt bodies (Letouzey et al., 1995), and extension during the early Mesozoic may have facilitated detachment folding through the aggregation of salt bodies within the basin. However, gravity, seismic, and stratigraphic data do not support the involvement of Triassic salt in the development of the Alima anticline. Due to the low density of salt with respect to other rock types, the presence of salt should produce a gravity low if diapiric doming produced the Alima anticline. Although gravity lows associated with anticlines north of the field area have been interpreted as a product of salt domes (Jallouli et al., 2005), the Alima anticline is associated with relative gravity highs, and the best fit forward model for the Alima anticline did not incorporate salt (Fig. 7). Furthermore, seismic-reflection lines in the Metlaoui region also do not suggest salt diapirism in the subsurface around the Alima anticline. Finally, well GNT-1, which penetrates 5 km of stratigraphy, shows no accumulation of salt of Jurassic age or younger. It is conceivable that salt involvement may have occurred in a manner similar to the “salt glacier” model proposed by Vila et al. (1994, 1996). In this model, salt is present not as an individual bed, but as a component of siliciclastic or carbonate mudstone, and the salt still acts as a zone of weakness. However, in all salt models of the Tunisian Atlas, Triassic salt is modeled above Early Cretaceous (Albian/Aptian) rocks. If this were the case for the Metlaoui region folds, then (1) some indication of salt in well GNT-1 should be present, which penetrates Jurassic rocks; (2) salt should be observed in the seismic data, which show stratigraphy down to the Early Cretaceous; and (3) the gravity signature of such a shallow, low-density feature would be apparent in both the gravity data and modeling. Based on the far-foreland nature of the stratigraphy, the gravity observations and forward model, and the well data, salt appears to have had limited involvement in fold development in the Metlaoui region.

Folds above E-W–striking faults farther south (i.e., Fault of Chotts) are interpreted to have formed from oblique reactivation of Mesozoic normal faults; the data in this study support a similar model for development of the Alima anticline and other folds in the Metlaoui region. Fold geometry in reactivation models includes tight, vergent anticlines and broad synclines, with en echelon, arcuate fold crests (Letouzey et al., 1995), i.e., the same geometry observed at the Alima anticline and surrounding synclines (Figs. 10 and 11). Theoretical and physical models have documented the evolution of folds in multiple orientations of reactivation, and both orthogonal and oblique reactivation can produce the observed fold geometry at Alima (Letouzey, 1990; Letouzey et al., 1990; Vially et al., 1994; Letouzey et al., 1995; Tindall and Davis, 1999). During orthogonal reactivation, new reverse faults typically develop (Letouzey, 1990; Letouzey et al., 1990, 1995; Outtani et al., 1995), which is not observed at the Alima anticline. For oblique reactivation, the optimal orientation for oblique reactivation (20°–40°) is similar to the angle between Cenozoic shortening (NW-SE) and some preexisting Mesozoic normal faults (E-W), and it does not predict the development of new thrusts. We therefore favor oblique reactivation over orthogonal reactivation.

The oblique reactivation model (Letouzey et al., 1995) also predicts en echelon, vergent folds with en echelon fault traces, tear faults, and sometimes second, tighter anticlines when multiple detachment surfaces are included in the model (Figs. 10 and 11). The fracture sets are inferred to have formed early in the folding history, parallel and perpendicular to the curved and segmented fold crests. Thus, the existence of arcuate and segmented fold crests is likely not the result of subsequent rotation or deformation. The direction of shortening in the southern Tunisian Atlas Mountains is NW-SE, and therefore the relative direction of elongation is NE-SW (Zargouni, 1986), orthogonal to arcuate faults in the Alima structure mapped in this study. In addition, the Alima anticline is composed of right-lateral, en echelon folds (i.e., the respective anticlines associated with Zimra, Zarrif, and Alima peaks), with arcuate fold axes offset by tear faults, as predicted by models of oblique-slip reactivation in the presence of detachment layers (Vially et al., 1994).

Direct evidence of an extensional structure being reactivated as a contractional structure is limited to the fault visible in the seismic line MTO-057 (Fig. 9D). The fault displays normal offset, suggesting extension, but the overlying strata are gently folded, suggesting that the fault has accommodated contraction. Since it is the overlying strata that have been folded, we interpret the fault to have first accommodated extension, and then contraction.

Regionally, the Alima anticline was produced in a convergent tectonic setting, but brittle deformation is locally expressed as normal faulting and extensional fracturing. Both normal faults and extensional fractures dominantly strike either subparallel to or perpendicular to fold crests. The patterns in brittle deformation seen on the Alima anticline are similar to brittle deformation patterns observed in the forelands of other mountain belts. Laramide uplifts in the western United States exhibit similar structural patterns (e.g., Engelder et al., 1997; Bergbauer and Pollard, 2004), and seismic data have revealed that Teapot Dome (a doubly plunging anticline associated with Laramide deformation) is asymmetric, with a gently dipping eastern limb and steeply dipping western limb (Cooper et al., 2006). Both the Alima anticline and Teapot Dome exhibit normal faulting along the periphery, commonly subparallel or perpendicular to the fold hinge, despite having formed in a contractional setting. Kinematic studies of fault populations associated with Laramide folds in the western United States also suggest that elongation, rather than shortening, is likely to occur parallel to the fold hinge, facilitating the development of fractures and normal faults perpendicular to the hinge (Wise and Obi, 1992; Narr, 1993; Varga, 1993). Thus, the normal faulting observed in the Upper Cretaceous upper member of the Berda formation is likely a product of extension in the outer (younger) strata of the anticline.

Fold History in the Metlaoui Region

Based on timing relations and deformation styles observed from seismic-reflection lines and outcrop data, folds of the southern Tunisian Atlas Mountains are classified into three categories (Fig. 12). The three categories correspond to episodes of contraction previously identified in the mid- to Late Cretaceous, throughout the Cenozoic, and occurring present-day (Ben Ferjani et al., 1990; Bobier et al., 1991; Bédir, 1995; Bouaziz et al., 2002). Folds associated with the Gafsa fault and the Fault of Chotts first developed during the mid- to Late Cretaceous and represent the first phase of folding (Fig. 12A). Both of these faults have been interpreted as reactivated basement faults, and they show characteristics unobserved in other folds in the region (Boukadi and Bédir, 1996; Swezey, 1996; Hlaiem, 1999). Seismic data for the Gafsa and Chott Range provide evidence for unconformities that are interpreted as having formed during episodes of uplift during the mid- to Late Cretaceous (Zargouni, 1986; Bédir, 1995; Bouaziz et al., 2002; Khomsi et al., 2009). Folds associated with the Gafsa fault also have NW-SE–trending fold hinges, rather than the ENE-WSW–trending hinges of the Metlaoui Range (i.e., Alima-style), and they are constrained to the immediate vicinity of the fault. The seismic data and fold hinge geometry, accompanied by the interpretation that the Gafsa fault and the Fault of Chotts initially formed as normal faults to accommodate N-S and NE-SW extension, suggest that these folds formed due to oblique reactivation in a transpressional setting (Zargouni, 1986; Bédir, 1995).

The second phase of folding includes the en echelon folds within the Moulares and Metlaoui Ranges (including the Alima anticline) and a fold in the Gafsa and Metlaoui basins (Fig. 9D). These folds are generally broad with a tighter anticline on their steeper southern flank and show early and late Cenozoic growth strata. Based on these growth strata relations in outcrop, three episodes of folding are distinguished within the second phase of folds. Oligocene through Miocene strata (Sehib and Beglia Formations) typically rest unconformably on Paleocene through Eocene strata (Swezey, 2009), indicating uplift during the Eocene. At the eastern end of the Moulares and Metlaoui Ranges, close to the Gafsa fault, the Sehib and Beglia Formations lie unconformably on the Upper Cretaceous Berda Formation and Mansour Group. These relations indicate that uplift was greater near the Gafsa fault during Eocene deformation and continued the growth of structures formed in the first episode of folding. Lower Miocene strata are unconformably overlain by upper Miocene and Pliocene strata, indicating uplift during the late Miocene.

The last episode of folding took place during the Quaternary and is ongoing. The direction of shortening during this last episode was oriented approximately NW-SE to NNW-SSE (Bouaziz et al., 2002). Based on unconformities, the latest episode of folding has reactivated folds from previous (both Late Cretaceous and Cenozoic) episodes of deformation (Dlala and Hfaiedh, 1993; Ahmadi et al., 2004). Most of the gentle folds in the Gafsa, Metlaoui, and Moulares Basins (Fig. 9C) developed during this last stage of folding. Quaternary alluvium is folded around these structures, indicating that the folds are presently active. It is likely that the folds formed by oblique-slip reactivation, based on the presence of steep, underlying reverse faults. In addition, the folds are similar in structural style to the folds developed during the Paleocene–Eocene and Miocene (i.e., Alima anticline-style folds).

The Alima anticline, which forms part of the Metlaoui Range, is a broad anticline with secondary fold development parallel to its short axis (Zarrif Peak anticline) and long axis (Alima Peak and Zimra Peak anticlines). Fracture orientations are parallel and perpendicular to the overall trend of the long axis of the fold, suggesting that the complex geometries and orientations of the folds within the Alima structure are a product of the style of folding. A gravity survey of the Alima anticline and seismic data from nearby folds indicate limited involvement of salt in the formation of the Alima anticline.

The data presented in this paper favor the model of oblique reactivation due to far-foreland deformation for the formation of the Alima anticline. Overall, there is evidence for several episodes of uplift and folding in the southern Tunisian Atlas Mountains during the Late Cretaceous, Cenozoic, and Quaternary, although not all folds were active during each episode of deformation. This model accounts for deformation on a regional scale as well as the more cryptic structures observed locally in seismic data and field data at the Alima anticline.

Simo and Tikoff thank M. Cooke. Tikoff thanks C. Swezey for introducing him to the area in 1994 and many subsequent discussions. Soussi and Gordon thank M. Bédir, S. Primm, J. Graham, P. Gill, and A. Fares for field work. We thank A. Masrouhi, C. Swezey, and R. Russo for providing constructive reviews. Compagnie de Phosphate de Gafsa is thanked for providing the seismic sections. This project was funded by ExxonMobil.