Evolution of Pleistocene travertine depositional system from terraced slope to fissure-ridge in a mixed travertine-alluvial succession (Jebel El Mida, Gafsa, southern Tunisia)

The Quaternary stratigraphic record of Jebel El Mida, composed of continental deposits, is a useful example of concomitant travertines and alluvial deposition in an extensional setting. Travertine deposition occurred in a faulted Pleistocene alluvial fan giving rise to seven (recognised) facies interfingering with five other alluvial ones. The travertine depositional events indicate a tectonically driven evolution from terraced slope (facies group FC1–FC6) to a travertine fissure ridge-type depositing phase (facies group of FC1–FC7). Interfingering between travertine and alluvial facies indicates the co-existence of adjacent and time-equivalent depositional environments. The travertine deposition resulted from deep origin hydrothermal fluids channelled along damaged rocks volumes associated to a regional fault system, named as the Gafsa Fault (GF). The travertine–terrigenous succession in Jebel El Mida highlights the major role played by the GF in controlling: (i) the hydrothermal fluid flow, still active as also indicated by the numerous thermal springs aligned along the fault zone; (ii) paleoflow directions, discharge locations, volume, rate and fluctuations of the water supply. The paleoclimatic correlation with adjacent localities reveals that, at that time, humid episodes could have contributed to the recharge of the hydrothermal system and to the deposition of alluvial sediments.


Introduction
Travertine is a thermogene continental carbonate deposit associated with thermal springs, and whose deposition is induced by CO 2 outggassing favoured by the fluids' pressure drop, turbulence of running water and microbial activity (Pentecost, 2005). Travertine deposits are relevant in providing helpful elements (i.e. paleoclimate, paleoecology of the spring system, tectonic setting, paleohydrology, paleotopography and thermal history) for geological reconstruction (Brogi, Capezzuoli, Aqué, Branca, & Voltaggio, 2010;Brogi et al., 2016;Chafetz & Folk, 1984;Ford & Pedley,1996;Gandin & Capezzuoli, 2008;Guo & Riding,1999;Pentecost, 2005). The term travertine is used to indicate a variety of deposits associated with thermal (generally > 30°C) and HCO3-enriched (>7 mmol/l) springs through mostly abiotic depositional processes . The travertine mineralogical composition is dominated by calcite and/or aragonite (dendritic, blades and acicular crystals) with higher (cm to m/year) sedimentation rates of regularly bedded to finely laminated fabrics (Chafetz & Folk, 1984;Pentecost, 2005). The depositional morphologies include multisymmetrical bodies such as mounds, slopes and ridges . They are distinguished from tufa by the abundance of coated gas bubbles and shrub lithofacies (Guo & Riding, 1999). The hydrological setting of travertine deposits is dominated by regular and permanent flow hydrodynamics with a strict relationship to tectonics (Hancock, Chalmers, Altunel, & Çakir, 1999). The travertine deposition is scarcely influenced by anthropogenic or climate-controlling factors .
The term tufa is used to indicate a dominantly biotic continental carbonate associated with cool fresh water streams and springs (Ford & Pedley, 1996;Pedley, 2009). The tufa mineralogical content is dominated by calcite (micrite and microsparite) and the absence of the aragonite. Its sedimentation rates, ranging from few mm to cm/y, are lower than those characterizing the travertine deposition. The most conspicuous character of the tufa deposits is the abundance (over 40%) of the primary porosity due to the richness of its biological content represented by micro and macrophytes (Ford & Pedley, 1996;Jones & Renaut, 2010). The depositional morphologies of tufa include axial-symmetrical bodies such as cascades, dams and barrages with a distinctive phytoherm lithofacies. Tufa deposition is mostly controlled by climate change and deeply influenced by anthropogenic interference in its deposition system . Although the terms travertine and tufa are used to categorize continental carbonates formed in very different environments, there are cases where travertine and tufa can deposit in the same depositional system: in fact, travertine is reported to occur in different depositional systems, including terraced slopes (Altunel & Hancock, 1993b;Capezzuoli et al., 2016), waterfalls and cascades (Chafetz & Folk, 1984;Pentecost, 2005), depression fills (Guo & Riding, 1998), fissure ridges (Altunel & Hancock, 1993a;Atabey, 2002;Brogi & Capezzuoli, 2009;Brogi, Capezzuoli, Alçiçek, & Gandin, 2014;Brogi et al., 2016;De Filippis et al., 2013;Guo & Riding, 1999;Selim & Yanik, 2009) and channels (Drysdale, 1999). These depositional systems could laterally interact with different environments (i.e. fluvial or lacustrine) favouring tufa and/or lacustrine carbonate deposition interacting with terrigenous sedimentation. Description of these articulated and laterally variable depositional systems is scarce in the literature and is reported only for restricted areas (cf. Croci, Porta, & Capezzuoli, 2015;Huerta et al., 2016;Martini & Capezzuoli, 2014;Viles, Taylor, Nicoll, & Neumann, 2007).
In order to contribute to the better understanding of the sedimentary and tectonic interplay in a depositional system characterized by the interaction between travertine and alluvial terrigenous deposition, a key area exposed nearby the Gafsa city, southern Tunisia ( Figure 1), has been analysed. These travertine deposits have been scarcely studied and few data have been published on their age, origin and depositional systems; these have been considered so far as continental carbonates and described as calcretes and dolocretes (Abdeljaouad, 1991;Gallala, Gaied, Essefi, & Montacer, 2010;Regaya, 2002). Furthermore, these deposits are interlayered with alluvial sediments in a relatively wide alluvial fan system.
Recently, the travertine deposits in the eastern part of Jebel El Mida along the strike-slip GF in southern Tunisia have been documented by Henchiri (2014a). Furthermore, Jebel El Mida travertine has only been attracted for its archaeological remains encrusted in the conglomeratic level and composed of lithic flakes, hand axes, burins and scrapers assigned to late Acheulean age (Collignon, 1887;Gobert, 1952;Vaufrey, 1933).
The study of Jebel El Mida travertine allowed us to reconstruct an articulated depositional system characterized by travertine formation along with alluvial deposits in which facies changes of terraced slope to fissure ridge took place. In addition, we emphasize the interaction between travertine and alluvial terrigenous deposition in an active extensional basin, where tectonic (mainly) and climatic factors played a fundamental role in controlling the evolution of this continental succession.

Location and hydrostratigraphy
The study area is located 1 km to the south of Gafsa City within the north-west-trending ranges of southern Tunisian Atlas (Figure 1(A)). It is bordered to the north by the Gafsa Fault (hereafter GF), and covers an area of 15000 m 2 with a maximum altitude of 328 m. The range of daily and annual temperatures shows high seasonal variation with irregular precipitation as typical of semiarid and arid regions. The average rainfall ranges from 100 to 250 mm/y. The drainage network consists of south-draining intermittent streams including Wadi Kabir, which join Wadi Sidi Aich at Gafsa to form Wadi Bayech.
The aquifer levels belonging to the main hydrostratigraphic units are shown in Figure 1(B). Here, the oldest aquifer unit is called 'the continental intercalaire CI' (Edmunds et al., 2003) and it has a potential reservoir thickness ranging from 200 to 1000 m (Castany, 1982). The CI aquifer is hosted in Lower Cretaceous continental (fluvio-deltaic) deposits, i.e. Boudinar and Sidi Aich formations, which constitute the main aquifer levels of the CI. Lithological facies of the CI-bearing formations are variable but dominated by detrital facies. The Boudinar Formation, Hauterevian to Barremian in age, comprises an alternating sequence of sandy and clayey layers with a thickness ranging from several to 60 m. The Sidi Aich Formation is composed of 50-m-thick feldspar-rich and white friable sands. The second hydrostratigraphic unit is called 'the Continental Terminal CT' hosted in the 1000-m-deep upper Cretaceous and Cenozoic formations (Edmunds et al., 2003). Lithologically, the CT-hosting formations are different with distinct three aquifer levels. The main level in Gafsa region is located within the carbonates and evaporates of Zebbag Formation (Upper Cenomanian) with a thickness ranging from 120 to 150 m for the dolomite and chalky levels and 300-400 m for the upper marly, clayey and gypsum horizons. The youngest aquifer unit belongs to the Mio-Pliocene Beglia and plio-Quaternary Segui formations. The Beglia Formation (Miocenelower Pliocene) deposited in a fluvial environment, composing coarse-grained sandstone intercalated with mudstone, showing variable thickness reaching 400 m near the region of Moulares to West of Gafsa. This formation is overlain by the upper Pliocene loess bearing Segui Formation (Mannaï-Tayech, 2009), consisting of clays and silts with 150-400-m-thick continental sequence, and it is considered as the base of the Quaternary sequence.
The GF was generally assumed to be an inherited structure from the Mesozoic rifting of the southern Tethyan margin (Gharbi et al., 2014). The middle segment of the GF represents the oldest portion of the stratigraphic succession and consists of Triassic evaporites cropping out along the major fault. These units were locally over thrust on younger successions along NW-SE and E-W-trending right-lateral oblique-slip faults (Hlaiem, 1999). Nevertheless, the GF fault is still active as testified by numerous earthquakes occurring in the Gafsa region (Figure 1(C)). Here, the seismically active zones are concentrated along the major N140°s trike-slip fault. Instrumental seismological data show that almost 50 seismic events were recorded nearby the GF since 1975; the majority of them (30) have M > 3 with the strongest instrumental record during 12/27/ 1985(Ben Ayed, 1986. Kinematic data collected in this study indicate a transpressional movement (Figure 1(C)) compatible with the regional kinematic regime detected in several previous studies (Gharbi et al., 2014). Nevertheless, on the northwest-dipping fault scarp affecting Pleistocene alluvial deposits (about 150-m long, 1-6-m high and trends 115°and dips about 70°NE) (Figure 2

Material and methods
Restricted by the exposure of the travertine-terrigenous succession in the field (Figure 4(A)), six representative sections were logged and described (Figure 4(B)) in which sedimentary structures, lithologic components, stratigraphic relationships and geometry of the rock strata were recorded in detail in order to define the fabric types, their lateral and vertical multi-scale variations and the depositional processes governing their deposition and the environment in which they were deposited.
Field and satellite photographs of an area approximately 12-km wide and comprehending the water gap between the Jebels (ranges) of Ben Younes and Orbata have been analysed in order to identify the sites of measurement of stratigraphic section and determine geomorphic features to characterise the geometry and kinematics of the Major Fault (GF).
The stratigraphic sections have been chosen on the basis of their stratigraphic position, structural features, exposure and accessibility. Each section was described, sampled and photographed in the field. Sections were sampled at 0.1 m intervals depending on the nature of the stratigraphic section and travertine content. Travertine and terrigenous detrital samples were sampled in the field for the preparation of thin sections and polished (travertine) rock slabs and the results were used to identify grain, matrix and cement percentage to aid in rock classification. Grain size estimates, discriminated in the field, were made with a visual grain size chart based upon the Wentworth scale (1922). Re-examination of representative samples in the laboratory with binocular microscope confirmed the general accuracy of the field determinations.
The mineralogical composition and the low-Mg chemical character of the travertines were carried out with an X-ray powder GBC diffractometer with Cu-Kα X-ray radiation at an accelerating voltage of 40 kV and electrical current of 20 mA, and Brucker Vector 22-based FTIR spectroscopy, respectively.

Lithofacies
First attempt on the definition of the tufa and the travertine in Gafsa region has been made by Henchiri (2014a), in which the entire architecture of the depositional units was not clearly defined (e.g. fissure ridge). In this work, and owing to a more clear vision of the deposits comprising both terraced slope and fissure ridge, morphologies have been described as travertine rather than tufa. Two travertine outcrops were recognised in Jebel El Mida. The first corresponds to remnants of terraced slope travertine in the western part of Jebel El Mida ( (E)). This latter, even though faulted and uplifted, it is still recognisable as well its entire morphology. The exposure of the studied succession in the field (Figure 3(F)) has permitted the description of six representative sections (Figure 3(G)). Seven travertine lithofacies (Table 1) were identified and analysed according to their fabric multi-scale characteristics. Their facies code and description were based on the terminology proposed by Chafetz and Folk (1984), Guo and Riding (1998), Jones and Renaut (2010) and Croci et al. (2015). Granular travertine lithofacies were classified according to the scheme of Dunham (1962). Floatstone refers when 10% of the particles are larger than 2 mm with matrix support and rudstone refers clastsupporting (Embry & Klovan, 1971). The 'boundstone' terminology was ascribed to certain travertine lithofacies with rigid micritic framework. Five alluvial fan lithofacies have been distinguished (Table 2).

Description
These facies consist of shrub-like and spherical structures (Figure 4(A) and (B)) composed of dentritic peloidal micrite nucleating and flouring either planar or undulated exposed surfaces developed on various substrates. They exhibit diverse branching and clumping morphologies ranging from upward expanding small rounded knobs (0.2-1.5 cm high) to domal lobes (1-3 cm high) that grew and draped pre-existent substrates represented by laminated boundstone, rafts and reed boundstone. In thin section (Figure 4(C)), the shrubs show irregular branching morphologies with discernible radiating and arborescent appearance.

Interpretation
The dentritic shrub boundstone is interpreted as bacterial in origin (Chafetz & Guidry, 1999). The development is restricted to chemically harsh, hot-water travertine deposits in terrace pools, depressions and fissure ridge (Chafetz & Guidry, 1999) and their morphologies are strongly influenced by the bacterial colonies (Chafetz & Folk, 1984;Chafetz & Guidry, 1999;Chafetz & Meredith, 1983). However, such carbonate constituents have been ascribed to abiogenic precipitation processes by Pentecost (2005).  broken and disrupted laminae with corrugation and undulation where laminae do not exactly match each other. Dense laminated boundstones consist of dense laminae of beige to light grey in colour. Layers of laminated boundstone average from 2 mm up to 1 cm in thickness and locally disrupted by isolate sub-spherical to lenticular 3-mm-1-cm-wide pores (Figure 4(D)).

Interpretation
These lithofacies are encountered in the spring thermal pools of either terraced slope or fissure ridge depositing systems or in the poorly drained portions of the spring (Ford & Pedley, 1996;Pentecost, 2005). It is ubiquitous as demonstrated in modern travertine depositing systems such as those in Rapolano Terme in Italy (Brogi & Capezzuoli, 2009;Chafetz & Folk, 1984;Guo & Riding, 1998) and in the Denizli basin in Turkey (Altunel & Hancock 1993a;De Filippis et al., 2013;Khatib et al., 2014), where cyanobacteria with their associated biofilms contribute to the development of laminae by acting as substrate for the calcite nucleation and the fixation of the CaCO 3 grains. In addition, the laminated structure can indicate periods of stability of their support, which once moved can lead to the disruption and the breaking of the laminae. The alternation of laminae and the variability of crystal shape are attributed to seasonal fluctuations of the spring water temperature, chemistry and flowing regime (Andrews & Brasier, 2005;Chafetz, Rush, & Utech, 1991).

Description
These facies consist of few microns to 3-mm-thick white to beige colour (Figure 4(H)) deposits. Their morphology varies from flat to wavy depending upon the morphology of their substrate. They range from widely to densely packed sheets of carbonate films preserving locally intraraft porosity. The fragmented and re-sedimented raft fragments often develop cemented grainstone/rudstone lenses ranging in thickness from 5 to 10 cm and with lateral extension varying from 1 to several metres. Raft boundstone often serves as a precipitation support for shrub boundstone.

Interpretation
Raft sheets and fragments are interpreted as being the result of calcite or aragonite precipitation in the surface of stagnant waters belonging to pond systems and pools of terraced slope and fissure ridge systems (Guo & Riding, 1998). Some authors ascribed them to ponds near the hydrothermal vents (Folk, Chafetz, & Tiezzi, 1985;Jones & Renaut, 2008). The development of raft grainstone/rudstone to floatstone is triggered by various mechanisms. The continuous growth of calcite or aragonite crystals in the upper and lower raft surfaces increases the density of rafts that collapse and sink to the bottom of the pool. Stagnant waters are the main condition for the raft precipitation which explains why

Description
They include stem and branch moulds and leaf imprints ( Figure 5(A)). The main part of stem moulds preserves their original upright position with heavy carbonate encrustation composed of alternating bands of porous and dense micrite with variable band numbers and thicknesses ranging from 6 to 8 bands and from 2-to 3-mm thick, respectively. Moulds of reed stem are also abundant and can be either horizontally crowded (layers of 50 cm) and encrusted.

Interpretation
Plants and prokaryotes are always present in the majority of spring systems where water is maintained with constant supply, composition and temperatures that do not exceed their degree of tolerance to survive (Jones & Renaut, 2010). According to Jones and Renaut (2010), plant moulds diversity, density and distribution can indicate changes in the temperature of spring waters (cooling gradient) towards plant-tolerated temperatures as they flow downstream in the relatively distal sites and are almost absent in the vicinity of the ridge fissures or spring vents.
4.6. Coated grains (FC6) 4.6.1. Description They are represented by grains with a nucleus coated by cortical laminae (Figure 5(B) and (C)). Their density varies from scattered to densely packed and their size ranges from 0.5 to 3 cm in diameter. Cut pisoids ( Figure 5(C)) show closely packed concentric to subconcentric cortical laminae consisting of lamina couplets of dense dark and porous light micrite ranging in number from 10 to 20 dark laminae and 15 to 30 light laminae. Some of the laminae are more visible than others due to their low content of dark pigments (Fe and Mn oxides).

Interpretation
Pisoids can develop in many different types of spring systems (Andrews & Brasier, 2005;Chafetz et al., 1991;Ford & Pedley, 1996;Guo & Riding, 1998Pentecost, 2005) and they have been ascribed to inorganic processes associated with bacterial mediation in pools that periodically agitate (Jones & Renaut, 2008). Some variations of pisoid shape are interpreted to be induced by the shape of their nucleus (Pentecost, 2005).

Description
These particular facies consist of calcified cylindrical insect larval tubes giving the rock an appearance of roughly porous fabrics ( Figure 5(D)). Tubes are arranged in layers (up to 5 cm thick) with the individual tubes oriented normal to the bedding, roughly parallel to each other and slightly narrower to the base of the bed. Tubes are 5-10-mm long and 3 mm in diameter ( Figure 5(E)). The upper cephalic ends of the tubes are open with remarkable funnel-like openings. Inner tube walls are lined with 0.1-0.5-mm-thick beige to white dense micrite rings encased in a more thick and porous white micrite envelop that constitutes the main tube framework.

Interpretation
These facies are interpreted as calcified insect larval retreats (Carthew, Taylor, & Drysdale, 2002;Drysdale, 1998Drysdale, , 1999Drysdale et al., 2003). Based on the overall dimensions of the tubes, their arrangement and the texture of the inner tube micrite ring are interpreted as Pyralidae (order: Lepidoptera) larval cases (Carthew et al., 2002;Drysdale, 1998). Pyralid larval cases are constructed by silken structures, in the form of tube-like (burrow) cases (Carthew et al., 2002). However, this is not only the silken structure in which pyralid insects construct their larval cases but also the marquee shelters (rectangular structures) are as widespread as the tube-like cases depending on the ecological behaviour of the insect. According to Carthew et al. (2002), these burrows may indicate a preferential hydraulic environment within the travertine depositional system represented by gently flowing water over the travertine masses where open ends of the tubes face the current direction of flowing waters. These cases are constructed using the salivary silk secretions of pyralid insects to bind the detritus, algae and small animal debris to form the tube framework (Benke, Van Arsdall, Gillespie, & Parrish, 1984).

Description
These facies consist of matrix-supported conglomerates with sub-angular to sub-rounded and poorly sorted cobble-boulder clasts (Figure 6(A)) (clast size ranging from few centimetres to 1 m) with disorganised fabric. Crude imbrications are common as well as major channellised bases and possibly large flute marks or sharp truncations. The conglomerate layers, with thickness ranging from 0.5 to 2 m and decreasing towards the S-SE direction, stand out as crudely bedded terraces. The total number of these conglomeratic bars within the outcrop succession varies from 15 to 20 bars representing normal and inverse grading. The fabrics of the coarser clasts are disorganised and rarely graded. The lithological composition of the clasts consists of <70% limonite-stained boulders. Elongate boulders are also common and aligned parallel to bedding and concentrated just above the basal portion of the beds. The matrix shows two main components: (1) a detrital matrix consisting of terrigenous silty sand and (2) locally massive micrite and microsparite (Figure 6(B)).

Interpretation
These conglomerates of poorly sorted, muddy matrix-supported gravel and chaotic fabrics are interpreted as a deposition by cohesive clast-rich debris flow (Blair, 1987;Miall, 1996). The remobilisation of debris flow by water action is evidenced by the presence of crude-stratified gravel. The development of an alluvial fan body by debris flow deposition is documented in ancient and modern examples (e.g. Bull, 1977). The debris flow is defined by Blair & McPherson, 1994 as a type of sediment-gravity flow composed of a mixture of gravel, sand, clay and a small amount of entrained water driven downslope by the force of gravity. The origin of the intergranular filling by micrite and/or microsparite is probably ascribed to deposition under subaquatic low energy conditions in alluvial plain shallow ponds (Croci et al., 2015).

Description
These facies consist of disorganised, poorly sorted, ungraded and matrix-to clast-supported pebble, cobble to boulder clasts ( Figure 6(C)). The clast size varies from few centimetres to 80 cm with sub-rounded to rounded shapes. The large boulders lie along the basal contacts and are spaced several metres away. Bedding thickness averages between 20 and 60 cm. Beds have irregular basal contacts with lens-shaped geometry with non-stratified and chaotic fabric. The matrix is composed of a mixture of the same matrix components seen in FT1 plus coarse sand-granule matrix. Facies FT2 interfingers and intercalates vertically and laterally with FT1 with remarkable abundance of cobble-sized clasts within a polymodal clast distribution. Locally, they can show heavily encrusted fractures with white calcite veins.

Interpretation
The sedimentary rock attributes of FT2 suggest a rapid deposition pattern induced by hyperconcentrated flood flow triggered by catastrophic flood events and/or heavy rainfall and cloud bursts (Lavigne & Suwa, 2004). According to (Nichols, 2005), hyperconcentrated flow has been recognised in instances where the catchment area was relatively small or where the hinterland bedrock was dominated by mudstones and evaporites. The fractures showing calcite encrustation are interpreted as conduits for carbonate-enriched fluids.

Stratified sandstones (FTS1) 4.10.1. Description
These facies FS1 are present locally at the base of the central logged section (Log WT3) beneath the disorganised matrix-to clast-supported conglomerates FT2 and consist of medium-to coarse-grained gray sandstones with clearly visible normally graded trough cross-stratifications ( Figure 6(D)). Set and coset thicknesses of crossstrata average between 20 and 35 cm and 1 and 1.5 m, respectively, with low-angle cross-strata dipping ranging from 15°to 25°. After removing the effects of dipping and tilting related to faulting, these cross-stratifications indicate a south to south-east paleocurrent direction. Relatively large-scale cross stratifications with 50-cm-high set are also identified with tangential, concave-up cross-strata. FTS1 passes upward and laterally into more massive and/or pink-coloured planar and laminated sandstones. Horizontal and thinly laminated beige-coloured fine-grained sandstone layers are separated by only millimetre-thick marls partings that weather out leaving behind them a fabric-oriented porosity.

Interpretation
Cross-stratifications are engendered by the downward migration of current ripples suggesting a relatively rapid deposition under waning stage of alluvial channel or by some other low-magnitude flow (Blair & McPherson, 1994;Miall, 1996).

Massive sandstones (FTS2)
4.11.1. Description These facies consist of fine to very fine-grained moderately consolidated massive sandstones (Figure 6(E)). The beds are laterally discontinuous and range in thickness between 50 cm and 1 m and are interbed with the disorganised matrix-to clast-supported conglomerates FT2. When they are overlain by FT2 and/or FT1, the transition is made through major channellised bases and possibly large flute casts and crude imbrications. Locally, these facies can contain prehistoric flint stone tools composed of lithic flakes, hand axes, burins and scrapers of late Acheulian age (Collignon, 1887;Gobert, 1952;Vaufrey, 1933) which are commonly dispersed with no preferential graded and/or oriented distribution.

Interpretation
This facies is interpreted as being deposited by rapid sedimentation through gravity-flow in shallow water ponds. Locally, they can represent the conglomeratic bar surface deposits of an alluvial floodplain (Miall, 1996).

Massive mudstones (FT3) 4.12.1. Description
These facies consist of green, reddish brown to yellowcoloured massive (to slightly laminated) and thickly bedded siltstones and claystones (Figure 6(F)) that occur throughout lower portions of entire logged sections. They show irregular contacts with overlaying beds. Locally, root cast and incipient pedogenic features (calcareous concretions with size ranging from 1 to 4 cm in diameter) can be identified randomly distributed with the reddish brown (Dry colour: 5YR 4/4reddish brown) claystones.

Interpretation
The fine-grained nature and the uniform texture of the siltstones and claystones are interpreted as over bank deposits (Blair & McPherson, 1994;Miall, 1996). The calcareous concretions indicate pedogenic activity and paleosol stage in the fan system Croci et al., 2015;Miall, 1996).

Terraced slope travertine lithofacies association
Terraced slope travertine lithofacies association is located in the lower portion of the travertine-terrigenous succession and recorded in the stratigraphic Log WT1 of the western compartment of Jebel El Mida. It comprises the facies group FC1-FC6 and shows clear massive and bedded structures with white greyish colour and weakly porous fabrics. These deposits are considered as the first generation of travertine (Figure 7). The travertine bedding plane surfaces show undulating structures with many protuberances and corrugations when restored to their original horizontal position by removing the tectonic tilting. They reveal much about their lateral continuity and the flowing direction of spring waters and even the distribution patterns of the overlying conglomerates. Figure 7(A)-(C) shows the general morphology of the travertine with the conglomerates, consisting of discernible pools of terraced slope system with rims and walls oriented to the south-east. Coarser boulders, cobbles and pebbles occupy the deepest and sheltered depressions of the terraced slope system in form of entrapments. The paleocurrent reading of fining upward detrital elements shows a south to south-east transport direction of the alluvial fan drainage.

Fissure ridge travertine lithofacies association
The fissure ridge travertine lithofacies association is recorded in the stratigraphic logs of the eastern compartment of Jebel El Mida (Logs ET1, ET2 ET3) and represented by the lateral and vertical transition of FC1-FC7. The uppermost denudated surfaces of the ridge (Figure 8(A)) can reveal small-scale ephemeral pools with rims (Figure 8(B)), with centimetres deep and less than a metre extend. Travertine discharge apron indicates the outlets of small springs within the fissure. The spring thermal pools show densely cemented rocks composed of shrubs of cynobacterial growth (FC1). In the thin sections, these rocks reveal a laminated structure (FC2). Many en-echelon and bifurcated fissures run sub-parallel along the ridge crest (Figure 8  overlain by coarse matrix-to clast-supported alluvial conglomerates (FT1 and FT2). The north-eastern parts of the ridge show accumulations of porous travertine around parasitic fissures which were developed on matrix-supported gravels (Figure 8(H)).

Alluvial fan lithofacies association
The visual alluvial fan deposits recorded in the six stratigraphic logs are distributed in paleostreams and interfingering in the eastern and western parts of Jebel Mida with travertine masses (Figure 9(A)). The alluvial fan lithofacies association is consisted of conglomerates and sandstones lithofacies (FT1, FT2, FTS1 FTS2 and FT3) ranging in thickness from 80 to 120 m. These deposits are arranged in fining upward sequences indicating waning flow conditions. The coarser fanglomerates with sandstone facies are developed as debris flows and streamflood deposits (Figure 9(B) and (C)). In Jebel El Mida, these alluvial deposits interfinger with and overlain by travertine deposits (Figure 9(D)).

Reconstruction of depositional events
Based on the vertical and lateral stratigraphic architecture of the travertine and terrigenous successions studied herein, two depositional phases can be distinguished for the travertine interfingering with alluvial deposits. A possible paleoenvironmental reconstruction of the different depositional phases is summarised in Figure 10.

Phase I
This phase (Figure 10(A)) is marked by the deposition of the western travertine at the base of the succession in the form of terraced slope mounds with a south to southeast dipping and thinning. This orientation indicates the location of the hydrothermal springs and the paleoflow direction in the immediate vicinity of the Gafsa Fault. The deposition of the travertine is assured by faultrelated hydrothermal springs suggesting a tectonic activity of the Gafsa Fault during early Pleistocene time (Henchiri, 2014a). This tectonic activity has provided conjugated faults for the travertine deposition and also the relief accommodating the alluvial fan deposition. The earlier NW-SE-trending dextral strike-slip faults with normal components linked to the Gafsa Fault have resulted in subsidence to create the depression where the alluvial fan deposited. The coarse clastic alluvial fan body is surrounded by fine-grained alluvial plain deposits with a limited number of individual alluvial fan apexes. Such spatial distribution of the alluvial fan facies  associations has been interpreted as the result of a drainage network that was expanding in the catchment area (hinterland drainage network) and localised between the structures of Jebel Ben Younes and Jebel El Assala. Overall, the alluvial fan sediments of Jebel El Mida record the history of sediment dispersal and deposition in sedimentary basin adjacent to a mountain belt. These alluvial sediments were formed by periodic flash-flood sedimentation in arid to semi-arid settings.

Phase II
This phase (Figure 10(B)) is characterised by the filling up of the former drainage network and the onset of clay-rich sandy alluvial/colluvial plain deposits. It is concomitant with a second travertine deposition phase that occurred through intermittently synsedimentary reactivated faults linked to the GF forming the fissure ridge depositing system. Co-existence of the travertine and the alluvial plain deposits during early stages of phase II is well materialised by the interfingering between both lithofacies in many places of the depositing system such as the travertine filling up of channellised streams and the sand-encased travertine blocks. The rapid growth of the travertine ridges has exceeded the slow alluvial plain development that resulted in the preservation of the ridges from flash flooding and coarse debris flow sedimentation which overlain only the low-elevated travertine masses in relatively distal sites. This depositional phase is associated with an upper Acheulean stone tools industry which was ascribed to early human (i.e. Homo habilis) by Collignon (1887), Vaufrey (1933) and Gobert (1952).

Chemical composition of the Gafsa Fault active springs
Hydrochemical parameters of water from some active springs along the Gafsa Fault near Jebel El Mida, including water temperature, pH, EC, pCO2, TDS, major and minor ions, were provided by the CRDA Gafsa (Commissiariat Régionale au Développement Agricole) ( Figure 11). The selected samples (from Ain Cherfa near Ben Younes and Ain Soltan near Orbata, see Figure 1) are characterised by a temperature averaging 30°C and groundwater pH values ranging between 6.76 and 8.1 (Figure 11(A)). The pCO2 ranges between 1.32 × 10 −2 and 5.38 × 10 −2 atm compared with the precipitation pCO2 of 10 −3.5 atm. The EC varies between 0.72 and 9.15 μS cm −1 . Higher values have been found near the GF (hydraulic sill). The total dissolved solids (TDS) range from 0.5 to 15 g l −1 (Figure 11(B)). The springs have very similar chemical composition (Figure 11(C)) which are characterised by a very high amount of SO 4 2− , Ca 2+ and Mg 2+ coupled with a relatively high amount of HCO 3− + CO3 2−. These springs have high contents of Sr, which, collectively sulphates and bicarbonate, indicate an interaction of fluids with a carbonate and evaporite succession.

Discussion
Hydrothermal activity along GF has been attributed to the result of the interaction between heat source, circulating fluids and permeability pathways (Saïd et al., 2011). The circulating hydrothermal fluids are of meteoric origin and were contaminated by a continuous water-rock interaction (Edmunds et al., 2003). The region of Gafsa is characterised by geothermal gradients ranging from 25 to 35°per km (Figure 11(D)) (Ben Dhia, 1990). This anomaly is mainly controlled by the NW-SE-trending strike-slip GF suggesting a strict control of such a structure in fluid flow circulation in a convective geothermal system. Deep water percolation and flow at depths of 1000 m or more (Lower Cretaceous 'Sidi Aïch Formation' 1000 m deep and the Upper Cretaceous 'Zebbag Formation' ≈ 340 m deep) increase the temperature of water by a variety of processes through which carbonate dissolution is enhanced by increase of water temperatures resulting from a deep-seated geothermal source since the study area is characterised by a relatively high local geothermal gradient and active tectonics (Ben Dhia, 1990). On migration to the surface, these CO 2 gasenriched solutions undergo rapid cooling and CO 2 loss, which shift the equilibrium position of calcium-bicarbonate groundwater to the travertine formation. Permeability pathways for these CO 2 −-enriched waters result from fault movement that fractures the rocks and, in turn, increase the overall permeability and thus generate new fluid paths to the surface (Caine, Evans, & Forster, 1996). Highly fractured rocks contribute sufficiently in the penetration of meteoric waters to deeper levels and mixing with other fluids of deep origin with the possible production of highly bicarbonate-enriched waters through reaction with carbonate rocks.
The travertine formation during lower to middle Pleistocene at Jebel El Mida implies the existence of paleohydroclimatic conditions characterised by abundant effective rainfall in the recharge zone. The second parameters are a dense vegetation cover, organic-rich soils that permit the percolation of acidic (aggressive) waters, the dissolution of the aquifer carbonate rocks and bicarbonate-enriched waters. It is noteworthy that the climatic, hydraulic and pedologic conditions that prevail during the travertine precipitation can be largely different from those prevailing during groundwater recharge. A gap between the two events is likely due to the speed of the water flow in the aquifer, which can only be estimated very roughly in ancient times and may be in some cases over last thousands of years (Marsily, 1981;Roche & Thiery, 1984). Radiocarbon dates of peaty material associated with continental carbonates (tufa) from Jebel Ben Younes (Henchiri, 2014b) to the west of Jebel El Mida reveal more younger deposits in the western parts of the Gafsa Fault. This may represent the continuity of the carbonate deposition from the springs that emerge from the fault after the uplift of the structure of Jebel El Mida and eventually dissect its travertine depositing system.
The U/Th radiometric dating of water and carbonates of land snail shells from Oued Akarit in Gabès region (95 km to the SE of Gafsa) shows the existence of four major rain phases and flooding events during middle to late Pleistocene (180-200, 130-150, 95-100 and 30 Ka BP) (Causse et al., 2003). In the Maknassy basin (55 km to the NE of Gafsa), the analysis of the carbonates of land snail shells dating from middle to late Pleistocene confirms the existence of several humid phases (Ouda, Zouari, Ben Ouezdou, Chkir, & Causse, 1993) more or less identical with those recorded in Oued Akarit. In the region of Chott El Gharsa (50 km to the SW of Gafsa), Swezey (1997) and Blum et al. (1998) have shown the existence of four lacustrine cycles (terraces) preserved within the series of incised valleys. Thermoluminescence dating of these terraces (Blum et al., 1998) gave ages of 111 Ka BP for the first, 108 ka BP for the second, 85 Ka BP for the third and 21 Ka BP for sub-recent terraces. These lacustrine cycles were characterised by the stabilisation of sand dunes of Chott El Gharsa by vegetation and raised groundwater levels. Correlatively in the region of Gafsa, which occupies the central location between these three neighbouring regions, it was inferred that there were same paleoclimatic events which contributed to sufficient recharge of the groundwater, the travertine development and the occurrence of alluvial deposition.

Conclusion
The morphotectonic and sedimentological study of a mixed travertine-alluvium succession at Jebel El Mida in southern Tunisia has led to the following conclusions: (1) The Quaternary stratigraphic record of Gafsa region shows at Jebel El Mida a mixed stratigraphic succession composed mainly of travertine and detrital terrigenous materials. Their distribution patterns and accumulation rates are function of tectonic and climatic variables.
(2) The travertine deposition takes place in a wide strike-slip-faulted and uplifted early to middle Pleistocene alluvial fan. Seven travertine facies (FC1-FC7) gave rise to terraced and fissure ridge-type travertine lithofacies associations, respectively. (3) The depositional events of travertine comprise a terraced slope travertine composed of the facies group FC1-FC6 which shows clear massive and bedded structures with white greyish colour and weakly porous fabrics bounded to the north-east by grainy travertine. Their deposition was interpreted to be triggered by inherited extensional conjugated fault system from the Gafsa Fault. The second depositional event is characterised by the onset of fissure ridge travertine composed of the facies group of FC1-FC7. Both depositional systems are separated in time and space and as interfingering with alluvial fan channels and alluvial plain deposits. (4) The hydrothermal activity feeding development of travertine at Jebel El Mida is a result of the interaction between heat source, circulating fluids and permeability pathways. The heat source is originated from a relatively high local geothermal gradient and active tectonics. The circulating fluids come from deep (1000 m) Aptian carbonateenriched groundwaters (Sidi Aich and Zebbag formations, respectively). The permeability pathways are provided by highly damaged rocks in the vicinity of the active strike-slip Gafsa Fault. (5) The paleoclimatic correlation with neighbouring localities reveals the existence of many humid episodes during early to middle Pleistocene. It has a possible contribution for sufficient recharge of the groundwater, the travertine development and the occurrence of alluvial deposition. This phenomenon has resulted in the stratigraphic record as the net sum of the tectonic activity, the rise and fall of the water table and associated climatic change.