Introduction

Stable isotopes in fluvial and lacustrine carbonates are useful as paleoclimate proxies (Leng and Marshall 2004) and carry other paleoenvironmental indications. High values of δ13C are infrequent in fluvial and lacustrine carbonates (Utrilla et al. 1998), and only those lacustrine carbonate facies linked to sulfate-poor waters may produce diagenetic carbonates in a methanogenic environment and/or in a highly productive lake resulting in positive δ13C values (Talbot and Kelts 1991). In continental carbonates, δ18O values are mainly dictated by the isotopic value of meteoric water and its temperature, while δ13C values reflect the origin of DIC (dissolved inorganic carbon) in the fluids and are largely determined by the relative proportion of isotopically light CO2 coming from soil organic matter and the heavier carbon sourced from the dissolution of aquifer limestones (Andrews 2006). The case study shows a carbonate complex ranging from calcrete through palustrine, lacustrine and, finally, travertine facies, allowing the analysis of contrasting facies within the depositional system filling the basin with expected differences in isotopic values.

The isotopic composition of soil carbonate (calcrete) is predominantly controlled by soil CO2, a mixture of atmospheric and respired CO2 (C3/C4 proportions of biomass), although other factors, such as depth of carbonate precipitation, production rate of soil CO2, temperature, absolute pressure, and atmospheric concentration can have an effect (Cerling 1991). The calcrete deposits show generally negative δ13C and δ18O values (Calvo et al. 1995; Driese and Mora 2002), and they are enriched by around 30‰ δ18O relative to soil water, and about 10‰ δ13C relative to soil CO2 (Friedman and O’Neil 1977). Cerling et al. (1989) reported that modern soil carbonates are systematically enriched by around 14.5‰ relative to organic matter in the soil; the diagenetic processes do not produce apparent changes on the original carbon-isotope signal (Cerling et al. 1989), although some studies indicate the alteration of the original δ13C values (Zhou and Chafetz 2009, and references therein). In calcrete accumulations, it is of great importance to know if parent material (both as allochthonous particles and as carbonate rock) occurs mixed with the carbonate precipitated in the soil. In this case, if we know the bulk isotopic composition of soil carbonate and that of the parent carbonate, the proportion of soil carbonate can be calculated (Salomons and Mook 1976).

Given that the palustrine carbonates result from a pedogenic modification of lacustrine carbonates (Freytet and Verrecchia 2002; Alonso-Zarza 2003), the range of δ13C and δ18O values is not always significantly different from that of the associated calcretes (Huerta and Armenteros 2005). As for the isotopic compositions of lacustrine carbonates, these are commonly used as paleoclimate proxies (Talbot 1990; Leng and Marshall 2004). δ18O is influenced by hydrologic balance and vapor exchange (Li and Ku 1997), whereas δ13C of DIC depends, in addition, on lake productivity (McKenzie 1984). In general, δ13C and δ18O co-vary in closed basins (Talbot 1990, and reference therein), and the carbonate lacustrine systems subjected to evaporation precipitate carbonates with higher δ18O values (Leng and Marshall 2004). A poor correlation of δ13C with δ18O values (r < 0.5) is characteristic of open lacustrine systems: small variations of δ18O and large changes in δ13C have been observed in some open lakes.

In many fossil examples, there is a close relationship between isotopic composition and facies (Arenas et al. 1997) and/or carbonate phase (Utrilla et al. 1998; Bowen et al. 2008). Travertine and tufa carbonates (after Ford and Pedley’s 1996 classification), or thermogene and meteogene travertines (after Pentecost’s 2005 classification), are terrestrial carbonates linked to springs and some specific fluvio-lacustrine systems, mainly differentiated by water source and temperature (Jones and Renaut 2010). Tufa deposits precipitate at low temperature (commonly less than 20 °C) and at low deposition rates, having δ13C values ranging between − 2 and − 12‰. Travertines, on the other hand, precipitate from warmer (20–40 °C), mesothermal (40–75 °C), and hyperthermal waters (> 75 °C) under higher pCO2 conditions and more rapid degassing, resulting in higher carbon-isotope ratios (− 2 to + 12‰) (Pentecost 2005; Jones and Renaut 2010).

The Borj Edouane Unit (BEU), located in the Quaternary El Gara Basin, hosts a singular framework of calcrete–palustrine and microbialite carbonate associations, asymmetrically distributed along the basin that allow us to reconstruct the depositional system. The sedimentological characterization of the carbonate facies and their associations (Ghannem et al. 2016; Ghannem 2018) have sparked our interest in investigating the geochemistry of the environmental setting under which they were deposited. Above all, this is an opportunity to document and analyze the relationship between the isotopic variability and the lateral sequence from calcrete–palustrine facies to microbialite facies, as represented by the change from fluvio-lacustrine stromatolites to travertine facies.

Current climate of inland North Tunisia is dry, and no lacustrine systems exist. The travertine and lacustrine carbonates of Middle–Late Pleistocene BEU indicate more humid conditions at the time of deposition. It is widely documented by the recent literature that there have been several humid periods of relatively short duration during otherwise rather arid Middle–Late Pleistocene time (Szabo et al. 1995; Crombie et al. 1997; Causse et al. 2003; Geyh and Thiedig 2008). Besides the chronological study of calcareous concretions and shells in detrital deposits of chotts in southern Tunisia (Causse et al. 1989), the only Pleistocene (Early–Middle) continental carbonate deposits known to us in Tunisia have been recently documented in relation to hydrothermal fluids and a humid episode in central Tunisia (Henchiri et al. 2017). Therefore, the research reported here represents the first documented carbonate system including travertine deposits in northern Tunisia that offers the possibility of reconstructing Quaternary climatic succession in the region. In addition, determining the characteristic isotopic imprint for the depositional system of the BEU basin can help to clarify the main constraints that control isotopic response throughout the system.

Geological setting

The continental carbonate Unit of Borj Edouane (BEU, Mid-Late Pleistocene) crops out in the El Gara Basin, northwest Tunisia (Ben Haj Ali et al. 1985). The El Gara Basin forms an elliptical plain between the Sarrath, Mellegue, and Ezarga Rivers, with an area of approximately 20 km2. Geologically, it is located at the transition between the central and northern Tunisian paleogeographic domains. The BEU occupies the central circular area of the Quaternary El Gara Basin, where it lies unconformably on the El Gara brachyanticline (axis oriented SW-NE) developed on Cretaceous rocks. In the area of BEU, the Cretaceous succession (upper Aptian and Albian in age) consists of carbonate units separated by marl-carbonate alternations (Gouasmia et al. 2006). The anticlinal structure is affected by a conjugate system of vertical faults with N–S and NW–SE orientation (Gouasmia et al. 2006; Ghannem 2018) that cross the eastern area of BEU (Fig. 1).

Fig. 1
figure 1

Geological sketch map of the studied area in El Gara Basin and location of the stratigraphic logs (asterisk). logs D and TJCB have been represented in Fig. 3, log U in Fig. 4 and log 18–25 in Fig. 5

Full size image

Methods

Detailed stratigraphic and sedimentological data have been gathered from more than 20 stratigraphic logs and many local outcrops. The petrographic study was done on more than 100 polished hand specimens and around 90 thin sections (including 9 polished sections for cathodoluminescence). Several samples, selected to represent the different facies identified as well as the regional Cretaceous carbonates hosting the BEU, were collected for isotopic analysis along the stratigraphic logs and from the outcrops.

δ13C and δ18O isotopic ratios were determined on CO2 liberated by acid treatment (103% H3PO4, under vacuum, at 25 °C; Craig 1957) from about 10 mg samples previously roasted on a plasma asher (to eliminate any organic content). Analyses were done on a dual-inlet SIRA-II mass spectrometer. Accuracy and average precision (± 0.02‰ for δ13C; ± 0.12‰ for δ18O) were monitored by repeated analysis of internal and international (NBS-19) reference materials. Results are reported in “delta” notation relative to V-PDB (δ13C) and V-SMOW (δ18O) in Table 1 and Fig. 6.

Table 1 Stable-isotope data of lacustrine–palustrine carbonates from BEU, Holocene Travertine, and underlying Cretaceous samples in the Gara Basin
Full size table

Sedimentology

Facies associations

The BEU is a carbonate unit that beyond the El Gara Basin passes laterally into detrital carbonate conglomerates and sandstones (Ghannem 2018). It has a maximum thickness of 12 m at its southern margin. In the western and northern areas, the unit mostly consists of tabular beds of massive calcretes, palustrine limestones, and laminar crusts, whereas in the central and eastern parts of the unit, microbialite facies are dominant (Fig. 2). The detrital facies crops out especially in the western, north, and northeastern margins. The alluvial association (sandstone and conglomerate facies, rich in carbonate clasts from the Cretaceous substrate) represents sedimentation in medial-distal alluvial fans. The BEU consists of two principal facies associations: (1) calcrete–palustrine–laminar crust association and (2) microbialite association (Ghannem et al. 2016). Within the microbialite association, there is a gradual transition from lacustrine stromatolite to travertine structures. In this paper, we follow the classification of Ford and Pedley (1996) for the differentiation of tufa from travertine deposits, which are, respectively, equivalent to those corresponding to meteogene and thermogene spring systems (Pentecost and Viles 1995; Pentecost 2005).

Fig. 2
figure 2

Map of facies distribution of the Borj Edouane Unit, with indication of samples collected and the position of the NW–SE section represented in Fig. 7. The different facies associations are arranged in roughly N–S parallel traverses

Full size image

Calcrete–palustrine association

The calcrete–palustrine association is mostly developed along the western and southwestern areas where it lies on alluvial facies in the marginal fringe outcrops. In the southern area, it is interdigited with the microbialite facies. Laminar crusts are widespread throughout the top of the BEU, representing overlapping secondary features, that are not associated with the underlying facies. Laminar crusts also occur interstratified within the microbialite association. In the northern area, the laminar crusts develop from conglomerates and calcrete–palustrine facies (Fig. 2).

Calcrete facies

The calcrete facies is represented by massive tabular beds up to 0.5 m thick, consisting of homogeneous micrite with silt-sized grains of quartz (1–20%), sand-sized lithoclasts, and fossils from the Cretaceous carbonates of the substrate. In some cases, the matrix consists of silt-sized micritic particles. They show channel, planar (crack-like), and vesicular porosity, and features common to alpha calcretes (sensu Alonso-Zarza and Wright 2010).

Palustrine facies

Sediments belonging to the palustrine facies form tabular beds from 0.15 to 0.80 m thick closely associated with calcrete facies and, toward the basin center, are interbedded with microbialite facies along a north–south boundary (Fig. 2). They consist of massive micrite, and contain silt-sized quartz grains (1–4%) and allogenic components (lithoclasts and fossils, and small Cretaceous planktonic foraminifera); scattered small oncoids occur. The presence of gastropod and ostracod remains is useful to differentiate the palustrine facies from the calcrete facies. Palustrine facies show abundant channel porosity (rootlets), are brecciated, and, occasionally, show secondary intraclastic–clotted–peloidal texture (Fig. 3).

Fig. 3
figure 3

a Stratigraphic log of calcrete–palustrine association at D. For symbols, see Fig. 4. b Outcrop at the point of D log. Note the nodular-to-massive structure and the intense mottling of the calcrete–palustrine transition. c Microphotograph of palustrine microfacies with micro-pelletized aspect of the micrite matrix; note ostracod valve (O) and fragments of shell molds of gastropods (G) filled with white sparite. Sample D1, plane-polarized light (PPL). d Stratigraphic log of the stromatolite association at TJCB site. e Field detail of the stratified structure of the stromatolite facies; note the horizontal alignment of the elliptic sections of oncoids showing intercalations of discontinuous planar stromatolite structures (arrowed). f Oncoids showing alternating white (sparite) and dark (micrite) microwavy laminae around nuclei (N) of microbialite gravels; oncoids are hosted in micrite matrix with fine stromatolitic fragments (SF); note a fine ostracod valve. Sample TJCB3, PPL

Full size image

Laminar crusts

Laminar crusts overlie calcrete and nodular palustrine facies and are stratified into tabular beds, forming strata 0.15–0.40 m thick, colored from light brown to pale orange. They can overly altered (calcretized) facies of alluvial-fan conglomerates.

The laminar crusts show a submillimeter corrugated lamination (600–800 µm thick) marked by the alternation of light and dark laminae. Crusts are associated with desiccation (mud-cracks and tepee structures) and brecciation. Dispersed gastropod fragments are present. Lighter laminae are thicker than darker ones and consist of a silt-sized clotted-peloidal texture richer in silt-sized quartz grains (≈ 1%). Dark laminae are micritic and commonly divided into discontinuous anastomosed branches showing irregular microdomes upward; their color is due to the vestiges of thin films of organic matter. These alternations are grouped into thicker ones (2–3 mm thick), where the darker levels are thinner. Laminated levels are commonly associated with intraclastic–peloidal levels (a few mm thick), where the elements are usually coated by a thin micritic coating of ordered clays.

Interpretation

The calcrete reflects carbonate precipitation in surficial soils developed on the distal fine-grained facies of the alluvial fans within shallow vadose-phreatic conditions (calcrete) (Huerta and Armenteros 2005). Calcrete deposits form a fringe that gradually transitions basinward to palustrine facies. The characteristics of the latter indicate sedimentation in shallow and low-gradient carbonate lakes most of whose area was affected by periodic floor exposure (Freytet and Verrecchia 2002; Alonso-Zarza 2003).

The textural and structural features indicate accretional formation in a palustrine domain (Freytet 1965; Vogt 1984), and probably reflect deposition in shallow restricted marginal puddles under dry conditions.

The microbialite association

Within this association three main facies are recognized: stromatolites, transitional travertine–stromatolites, and travertine microbialites. These facies lie in gradual transition from west to east (Fig. 2).

Stromatolites (oncoids and planar stromatolites)

This facies forms a north–south strip between the palustrine and the transitional travertine–stromatolite facies. It is mainly formed by oncoids with thin intercalations of wavy planar stromatolites. It consists of brownish orange, well lithified, tabular-to-wedge-shaped beds, 0.1–0.6 m thick, that can be grouped in sequences up to 1 m thick, where beds are bounded by well-defined erosion surfaces. Massive beds are common in northern outcrops where they are associated with the underlying alluvial conglomerates. Alternatively, the stromatolite facies may be represented by small, 1–2 cm fining-upward sequences displaying thick horizontal stratification, and, in some cases, cross and wavy stratification (Fig. 3). Oncoid beds, laminar stromatolite facies, and intraclastic microbialite levels are commonly interbedded. The oncoid beds include varied nuclei (intraclasts, mollusk bioclasts, and microbialite fragments), that can be multiple (polynuclear); their sizes range from submillimetres to a few millimeters, and show various morphologies (elongated, subspherical, and complex botryoidal-to-pustular/mammellonar shapes). Both oncoid and planar stromatolite structures show a characteristic submillimetric alternation of dark micritic and light sparitic laminae. Commonly, this lamination is included in vertically elongated sparite crystals as if they were formed by a poikilitic growth (sensu Friedman 1965). Oncoid texture ranges from poorly to moderately sorted. Two main textures have been recognized: (1) oncoids supported by an intraclastic-peloidal micritic mass (≈ oncomicrite of Flügel 1982) consisting of small oncoids, fragments of microbialites, peloids, and intraclasts with dispersed gastropods, ostracods (in some cases with articulated valves), and charophyte remains, that are patchily cemented by sparite; (2) oncoids within a scarce intraclastic-peloidal matrix and cemented by sparite (oncosparite); some levels show blue-dyed oncoids due to the content of Fe-rich calcite. Silt-sized quartz grains range from 2 to 10%.

Travertine–stromatolite transition

This facies shares features with the stromatolite and the travertine-like microbialite facies between which it is situated (Fig. 2) and forms beds from 0.10 to 0.45 m thick, organized in sections up to 1.7 m thick. Beds commonly show horizontal lamination (Fig. 4a–c). Intraclastic levels up to 0.6 m thick, with marked basal erosion surfaces, are common. Textures include a wide range; the more common being microbialite clasts, bacterial shrubs, and white amoeboid, botryoidal-to-oncoidal forms (0.5–1 mm in diameter) that are organized in different millimetric alternations 2–20 mm thick that make up the observed laminated beds. In both massive and laminated beds, interspersed micro-thrombolitic textures (sensu James and Jones 2016) are common.

Fig. 4
figure 4

a Stratigraphic log of the microbialite association represented by the transition travertine–stromatolite and travertine facies at U site. Note the intercalation of clastic microbialite beds on erosion surfaces. b Field aspect of the stratigraphic log U. Note the internal disconformities (D) underlying the clastic beds. c Peloidal–intraclastic micrite of the transition travertine–stromatolite facies; some micobialite fragments (Mb) are identified; note thin micrite coatings around particles. Interparticle spaces are occupied by light sparite mosaic; note some elliptic-to-circular porosity probably due to burrowing (B). Sample 27-3, PPL

Full size image

Travertine microbialites

This facies is named from its similarity with facies categorized in travertine settings (Chafetz and Folk 1984; Chafetz and Guidry 1999; Pentecost 2005; Jones and Renaut 2010; Gandin and Capezzuoli 2014; among others). Its outcrops are found in the southeast margin of the BEU, where they form tabular, well-laminated levels 0.75–1.5 m thick, in sequences up to 3.5 m thick (Fig. 5a). This facies occupies the highest altitudes of the basin and its surroundings, between a maximum of 493 m (in the eastern area, around OT outcrop near the fault system; Fig. 2) and approximately 475 m. The maximum dip (2.6% NW) is toward the center of the basin. However, in the opposite direction (southward from the area of maximum elevation), there is a terracing of discontinuous outcrops sloping (≈ 3.5 to 4%) towards the SSE, SE, and ESE, descending down to 445 m, toward Ezarga River. This facies forms beds 10–20 cm thick that show well-marked horizontal to low-angle lamination (Fig. 5b). Laminae are wrinkled at centimetric and millimetric scales (Fig. 5b, c) and, in some cases, form irregular domes. Some outcrops show bed sets with gently inclined foresets (Fig. 5b). Under the microscope, decimicron-to-centimicron alternations of dark micrite and light sparite lamina are of variable thickness, that commonly show wavy to zigzag tracing (Fig. 5c, e). Commonly, the lamination consists of dark micrite masses that form horizontal strips, botryoidal aggregates, and upright shrub-, bushy-, and fan-like arrays of filaments (≈ 10 μm wide), similar to bacterial shrub fabrics (cf Chafetz and Guidry 1999) (Fig. 5d). Other laminated levels are composed of palisades of blade crystals perpendicular to lamination (Fig. 5e) that can include horizontal thin dark micrite laminae and/or erect shrub-like micrite forms similar to crystal shrubs (cf Chafetz and Guidry 1999). Oncoidal, irregular oncoids, and cabbage-like structures up to few millimeters in diameter commonly form part of the lamination. Frequently, the above structures are absent or poorly preserved in some layers due to their replacement by a secondary sparite mosaic of non-equigranular anhedral crystals. Some travertine-like levels in the southeasternmost area of the unit show discontinuous decimicron plate-like layers formed by a thin micrite laminae coated by dentate sparite palisades that can grow both upward and downward, and can be distorted and broken (Fig. 5e). These resemble the raft structures common in travertines (Chafetz and Folk 1984).

Fig. 5
figure 5

a Stratigraphic log of the travertine facies at site 18–25 (see Fig. 2 for location). b Field aspect of stratigraphic log 18–25 (white dotted line). Note the marked lamination of the beds, with irregular wavy tracing (dashed-outline rectangles and insets) and the boundary interlayers (long dashed lines). c Alternations of dark micrite and light sparite laminae; many laminae are broken and collapsed and the continuous ones show wavy to zigzag tracing; partial cementation of the residual porosity. Sample ST, PPL. d Detailed field aspect of travertine facies with white irregular wavy laminae formed by lateral union of bacterial shrub-like forms (arrowed and enlarged inset). In the center left (La), two laminated levels appear similar to the laminar alternation represented in the lower part of e; dotted line outlines an internal cavity formed by partial destruction of lamination, and filled with intraclasts detached from the walls. Site 18–29, Fig. 2. e Detail of D. Fine alternation of microwavy micritic and sparitic laminae in the lower part; note the bladed crystals, perpendicular to and including the lamination (dotted outline rectangles mark crystals in extinction position). Above, thin micrite flakes (arrowed) interspersed with sparite laminae; dentate crystals (Dc) grow downward underneath some flakes (raft structures). Cross-polarized light. Note the similarity of b, c, and e with ‘cruststone-microbial, flat-to-curled laminites and puff pastry-like fabric’ of microbialites in thermal spring systems, as defined by Gandin and Capezzuoli (2014; their Fig. 8C, D and E)

Full size image

Interpretation

The main elements and structures of the planar stromatolites and oncoids are attributed to the binding, trapping, and precipitation of sediment by the extracellular polymeric matrix produced by microbial metabolic activities dominated by cyanobacteria (Burne and Moore 1987; Freytet and Verrecchia 1998; Riding 2000). On the other hand, the travertine structures are characteristically dominated by the early cementation on non-skeletal organisms (Riding 1992) that, in thermal systems, consist of a varied association of organisms, among which fungi, cyanobacteria, and photosynthetic and heterotrophic bacteria occur (Pentecost 2005; James and Jones 2016).

The well-stratified, oncosparite facies suggests deposition from currents, although oncoids do not need movement to grow (Golubic and Fischer 1975; Risacher and Eugster 1979; Jones and Renaut 1994; Hägele et al. 2006). The common coexistence of oncoids with microbialite clasts and intraclasts would indicate erosion and reworking by exceptional flooding events due to periods of increasing alluvial-fan discharge, which resulted in deposition in transitional fluvial–lacustrine and marginal lacustrine areas, especially in the northern part of the basin. In spite of the smaller oncoidal forms, these features resemble the recent formation of fluvio-lacustrine oncoids in the neighborhood of the Rhine entry into lake Constance (Schäfer and Stapf 1978). In contrast, the massive oncomicrite facies, that includes poorly sorted oncoids within a peloidal matrix and is widespread in the central lake basin, represents calmer lacustrine conditions. This interpretation is reinforced by the presence of limnic fossils (ostracods, in some cases with articulated valves, gastropods, and charophytes) and by the absence of exposure features; shallow water conditions are indicated by the absence of lamination and the presence of early diagenetic mosaics of Fe-rich calcite attributed to reducing conditions. The relatively abundant silt-sized quartz grains (up to 10%) and mesozoic lithoclasts indicates an abundant water supply to the basin, suggesting steady lacustrine conditions.

The planar stromatolites, which have similar lamination to that observed in well-developed oncoids, are attributed to the activity of cyanobacteria, by comparison with the classical types (Monty 1976; Burne and Moore 1987). The close relationship with the stratified oncolites with which they are interbedded (Fig. 3d, e), and the scarce exposure features seem to indicate a shallow lacustrine environment, probably deposited in low-gradient marginal areas of the lake (Platt and Wright 1991).

As is common in tufa and, especially, travertine deposits, the original micrite precipitates of bacterial origin are partially or completely substituted by sparite (Chafetz and Guidry 1999). The preserved bacterial shrubs in travertine systems are attributed to evaporation and decline in water level, whereas the intervening micrite laminae may reflect the settling of lime-mud following flooding conditions (Gandin and Capezzuoli 2014). Other interpretations for these dense micritic laminae linked their formation to the weakened carbonate deposition during the cold season (Chafetz and Folk 1984).

The outcrop structure of the travertine microbialites as revealed by the geographical facies distribution in the basin (Fig. 2), slight inclination of bed sets (Fig. 5b), disposition of outcrops along slopes of 2–4% around the highest altitude in the southeast area, and the close relationship with a system of vertical faults (Figs. 1, 2) suggest that their formation is linked to the presence of springs. Sheet streams flowing from the springs to the NW nourished the basin and, in the opposite direction, formed deposits on terraced levels down toward the paleovalley of the Ezarga River to the SE (Fig. 2). These conclusions are reinforced by the lateral change from the travertine structures to stromatolite and oncolite facies basinward, which is supported by the existence of a transitional lacustrine–travertine strip where the travertinic and oncoidal facies are mixed and form a bridge between the marginal areas of travertine development and the central basin, where oncolites and subordinately planar stromatolites occur (see Fig. 2). Furthermore, the microstructural features of travertine microbialites, such as bacterial shrub fabrics, abundant sparite cements, and raft-like platelets, point to a travertine environment (cf Jones and Renaut 2010; Gandin and Capezzuoli 2014).

Stable isotopes

δ13C and δ18O values measured are listed on Table 1 and have been plotted in Fig. 6. δ18O values of the lacustrine stromatolites are quite similar to each other (δ18OPDB = − 5.3 ± 0.2‰; n = 3; max = − 5.1‰, min = − 5.6‰), in contrast with δ13C values (δ13CPDB = − 3.6 ± 6.5‰; n = 3; max = + 7.5‰, min = − 3.9‰), which are highly variable. An obvious outlier, corresponding to a reworked oncolite, yielded δ13CPDB = + 11.7‰, δ18OPDB = − 2.6‰, which is remarkably similar to that measured on a Holocene tufa (δ18OPDB = − 2.8‰; δ13CPDB = + 12.5‰). Samples attributed to the calcrete–palustrine facies yielded values similar to those for the microbialite facies, with similar δ18O values (δ18OPDB = − 5.6 ± 0.6‰; n = 8; max = − 4.5‰, min = − 6.4‰) and more variable δ13C (δ13CPDB = − 0.1 ± 2.3‰; n = 8; max = + 2.9‰, min = − 2.6‰). The travertines yield δ18O values quite similar to each other, around − 6‰ (δ18OPDB = − 6.0 ± 0.3‰; n = 11; max = − 5.5‰, min = − 6.5‰). δ13C is, however, much more variable, with all values positive (δ13CPDB = 4.8 ± 2.7‰; n = 10; max = + 10.2‰, min = + 1.7‰). An obvious outlier, down to δ13CPDB = − 8.0‰, although with δ18OPDB in line with the remaining travertine samples, shows field and petrographic evidence of calcrete alteration. In contrast, the Cretaceous basement (bulk rock) has similar δ13CPDB (+ 1.5 ± 0.5‰; n = 4) but bimodal δ18O distribution, with three values within δ18OPDB = − 6.6 ± 0.2‰ and a single one at δ18OPDB = − 2.8‰. The carbonate cement of an alluvial conglomerate shows similar δ18OPDB (− 5.15) and a slightly positive δ13CPDB value (+ 1.05‰), very close to the majority of basement samples. No obvious correlations are suggested by the plot of Fig. 6. If anything, there is a hint of a trend of increasing δ13C with increasing δ18O of travertine samples, while the opposite (decreasing δ13C with increasing δ18O) seems to be the case for the calcrete–palustrine facies, although none of these is statistically significant.

Fig. 6
figure 6

δ13C versus δ18O plot of the carbonates analyzed. Note the variation in δ18O of the Cretaceous host limestones, without significant change of δ13C. The limestone with highest δ18O and δ13C is assumed to represent a primary signature, while that with lowest values is the one that most intensely interacted with meteoric (paleo)waters. From this starting point, the trend towards higher δ13C in travertine samples is interpreted to result from CO2 degassing and/or preferential use of the light C isotope during photosynthesis, while that toward lower δ13C in palustrine samples is indicative of increasing contribution of C derived from soil respiration. The hint of slightly higher δ18O values in either case reflects progressive evaporation of the waters in equilibrium with the most evolved carbonates

Full size image

Discussion

The palustrine–microbialite depositional system

The above interpretations of facies associations and the distribution of facies through the basin (Fig. 2) indicate the persistence of shallow, intermittent carbonate lakes along a north–south strip in the western area of the basin. This area is the domain of the calcrete–palustrine–laminar crust association that interdigitates and onlaps the middle-distal clastic facies of the alluvial fans sourced in the western margins of the basin. A transitional evolution from the alluvial to the palustrine facies occurs through a calcrete fringe that developed in areas of low sedimentation rate and was affected by the influence of shallow groundwaters derived from the surface (Armenteros et al. 1995; Sanz et al. 1995). Basinward (distally eastward), groundwater rose to the surface, resulting in the palustrine domain with intermittent ponding periods. This is the typical lateral relationship recorded in many alluvial-fan–palustrine systems (Alonso-Zarza 2003; Aziz et al. 2003; Huerta and Armenteros 2005). The sedimentological features (desiccation structures, tepees, laminar crusts, etc) of the carbonate facies indicate that water supply was intermittent and limited in this western area of the basin. The calcrete–palustrine facies changes eastward into a contrasting environment dominated by microbialite development in different sub-environments arrayed predominantly along the eastern margin of the basin. The environments where planar stromatolites and oncolites occur were situated toward the basin center, under fluvio-lacustrine conditions that required more permanent water conditions. In fact, the presence of similar stromatolitic structures has been interpreted as indicators of stable high lake levels during Late Pleistocene time in Lake Manyara (Casanova and Hillaire-Marcel 1992a). Commonly, these carbonate microbialite structures are formed in fluvial, lacustrine, and tufa-travertine environments (Ordóñez and García del Cura 1983; Pentecost 2005; Renaut and Gierlowski-Kordesch 2010). The change from fluvio-lacustrine microbialite structures into travertine facies toward the eastern area of the basin (Fig. 2) suggests the existence of springs draining toward the basin centre. In addition to the structural and textural features described above, this relationship is supported by the higher altitude of this area, where the Cretaceous basement is affected by a system of intertwined vertical faults (Burollet and Sainfeled 1956; Chihaoui et al. 2010; Jaillard et al. 2013; Ben Haj Ali et al. 1985; Gouasmia et al. 2006). In summary, through a west–east direction, there is a change from the calcrete and palustrine to travertine environments, passing through a fluvio-lacustrine environment that forms a central strip along the north–south direction (Figs. 2, 7). The characteristics of the corresponding facies allow the inference of an asymmetrical supply of water to the basin, so that the basin was progressively wetter toward the east, consistent with the interpretation that the dominant travertine facies in the eastern margin were linked to the springs mentioned above.

Fig. 7
figure 7

Conceptual model of facies relations along an NW–SE cross section of BEU (see Fig. 2 for situation; other symbols as in Fig. 4; J, TJCB, Un and 18–25 are situation of stratigraphic logs)

Full size image

Isotope geochemistry

In carbonates, δ18O values are mainly dictated by the isotopic value of water and its temperature, while δ13C values reflect the origin of DIC in the fluids, and are largely determined by inheritance and organic processes. A poor correlation of δ13C with δ18O values (r < 0.5) is characteristic of open lacustrine systems (Talbot 1990; Li and Ku 1997): small variations of δ18O and large changes in δ13C have been observed in some open lakes (e.g., Lake Huleh in Israel, discussed in Talbot 1990). High values of δ13C (higher than typical in marine carbonate rocks, > 3 to 5‰) are infrequent in fluvial and lacustrine carbonates (Anadón and Utrilla 1993; Calvo et al. 1995; Clark and Fritz 1999, and references therein; Arenas et al. 2000; Huerta and Armenteros 2005; Andrews 2006; Deocampo 2010, among others), and only those lacustrine carbonate facies linked to sulfate-poor waters may produce diagenetic carbonates in a methanogenic environment and/or in a highly productive lake resulting in positive δ13C values (Talbot and Kelts 1991; Leng et al. 1999).

A possible significant influence in some samples with negative δ13C values could come from the oxidation of organic matter. Respiration of soils vegetated by C4 plants (δ13C typically − 10 to − 16‰, with a mean value about − 12.5‰) and C3 plants (δ13C ranging between − 24 and − 30‰, with an average around − 27‰; Deines 1980; Voguel 1993) produces CO2 with low 13C/12C ratios. Dissolution of this CO2 in water, at the circumneutral pH values common in groundwaters and at environmental temperatures, produces HCO3− that, if in equilibrium with CO2 gas, has δ13C values ≈ 9‰ higher than those from CO2 (Clark and Fritz 1999).

Carbon isotopes

Within the paleogeographical context of El Gara Basin, ecosystems were likely dominated by C4 vegetation. At the Middle–Late Pleistocene time of formation of BEU carbonates, however, several humid episodes have been reported for the North of Africa (Szabo et al. 1995; Dearing et al. 2001; Rohling et al. 2002; Geyh and Thiedig 2008; Grant et al. 2017, among others), and there is evidence for at least three periods within the last 192 ka when the central Sahara sustained C3 plants (likely trees; Castañeda et al. 2009), although C4 photosynthesis was, admittedly, dominant. Notwithstanding the above, Dearing et al. (2001) provide evidence for Mediterranean woodland as far south as Tachine, some 300 km SE of the present location of BEU. Assuming maximum total fractionation of ≈ + 9.5‰ between soil respired gases and calcite at 25 °C predicts δ13C values for calcite between 0 and 5‰ for C4-derived CO2 (Cerling and Quade 1993), or about 10‰ lower if soil gas is derived mostly from C3 vegetation.

However, δ13C values around 0‰ in Quaternary deposits of Lake Akgöl have been attributed to the dominance of exchange with atmospheric CO2 (Leng et al. 1999). If average temperatures were somewhat lower, δ13C values in the range 3–4‰ would result in calcite precipitated from DIC equilibrated with atmospheric CO2 (Hillaire-Marcel et al. 1986; Hillaire-Marcel and Casanova 1987; Casanova and Hillaire-Marcel 1992a), considering that at 10 °C calcite precipitates at + 3.5‰ (Leng et al. 1999). In Pleistocene stromatolites of the African Rift lakes, high δ13C up to 5‰ has been attributed to the preferential use of light carbon by microbes during photosynthesis, which would cause a relative 13C-enrichment of the surrounding DIC and subsequent precipitation of 13C-enriched calcite (Casanova and Hillaire-Marcel 1992a). However, considering that the sedimentological characteristics of the Borj Edouane Unit (BEU) of El Gara Basin point to an open system in the eastern sector with HCO3− likely dominated by dissolution of the Cretaceous host and palustrine facies in the west, pointing to more locally restricted conditions, dominated by soil-CO2 (likely 1–2 orders of magnitude more abundant than atmospheric CO2), equilibrium with atmospheric CO2 becomes unlikely.

In addition, and since the BEU sits on top of Cretaceous carbonates, a potentially important, probably dominant HCO3− contribution derived from dissolution of the Cretaceous basement must be considered (Andrews et al. 1997; Leng et al. 1999).

From Table 1 and Fig. 6, it follows that δ13C values of Cretaceous limestones in the area are quite similar to each other at around + 1.5‰, while δ18O varies between ~ − 3 and ~ − 6.5‰. Measured values for the alluvial conglomerates, essentially eroded and transported basement limestones, have similar δ13C and intermediate δ18O values (see Fig. 6). Uppermost Cretaceous carbonates from the Tethys realm are characterized by slightly positive δ13C (~ + 1/+ 3‰) and slightly negative δ18O (~ −1.5/− 3.5‰) (i.e., data in Jarvis et al. 2002; Elkhazri et al. 2013), so that it would seem that sample CRT 3 (δ13C = + 1.9‰; δ18O = − 2.8‰; Table 1) represents primary sedimentary carbonate, while samples CRT 0, 1 and 2, together with CNG, with similar δ13C but lower δ18O, are the product of interaction (weathering) with meteoric waters.

It is highly likely, therefore, that DIC in the waters draining into Gara Basin had isotopic values not very different from δ13C = + 1.5‰; δ18O = − 7.0‰. Evaporation of the standing water mass, likely coupled with a rise of its temperature, favored CO2 degassing, and precipitation of lacustrine carbonates. Some samples with δ13C lower than the postulated DIC of the water, but similar δ18O (i.e., OT 3, C 1), likely record an increased contribution of respired CO2. The high isotopic ratios of C in sample TJCB 3, together with Fe-carbonate cements evidenced by staining, hint at precipitation under (somewhat) reducing conditions.

Samples of the calcrete/palustrine facies also display a component equivalent to DIC, as described above. In the palustrine environment, however, the relative contribution of run-off as compared with in situ processes is lower. Evaporation is also a factor, resulting in progressively higher δ18O, but DIC is no longer dominated by carbonate dissolution, and an important contribution of isotopically light, respired CO2 becomes evident, driving δ13C towards lighter values (Fig. 6). Obviously, the less water (higher evaporation and higher δ18O), the larger the proportional contribution of respired CO2, and therefore, the lower the δ13C. From the isotopic viewpoint, calcretes and laminar calcretes show no major differences.

Therefore, in terms of δ13C, values from 0 to − 6‰ can be expected for DIC, and from 1 to − 5‰ for the calcite precipitated from it at equilibrium. In this way, the lower values obtained from BEU (in the range − 3 to 1‰) indicate a possibly significant source of C from soil respiration. Therefore, the moderately lower and higher δ13C values suggest a variable balance between the relative contributions of C coming from oxidation of organic matter and carbonate dissolution in the catchment.

Higher values (> 4‰) in the BEU calcites may indicate, in addition, the participation of photosynthetic plants within a highly productive ecosystem where biomass was able to extract large amounts of CO2, leaving the aqueous HCO3− pool enriched in 13C, as suggested by Leng et al. (1999) and Causse et al. (2003). Some authors alternatively propose the effects of methanogenesis during a phase of low lake level. Methanogenesis forms bicarbonate ions enriched in 13C. However, the influence of organic methanogenesis cannot be proved in the fluvio-lacustrine system of BEU, although sparite mosaics linked to both cement and neomorphic fabric in some microbialite levels indicate reducing conditions (blue-dyed Fe-rich calcite after Dickson’s stain). Additionally, the possibility exists, based on recent fractures/faults observed in the Cretaceous substratum, that fluids coming from a process of thermocatalytic methane production deeper in the substratum could have contributed isotopically enriched DIC to the surficial water, resulting in the precipitation of 13C-enriched calcites. The tufa sample (TUFF, Table 1), with δ13C > 12‰ and comparatively high δ18O, and the reworked oncoid TJCB 3, with similar values (see Fig. 6), could be a hint in this direction. However, the participation of deeper-seated fluids of such nature is normally associated with a change in mineralogy due to dissolution/re-precipitation (Leng et al. 1999), an effect that is not proved here. Geophysical and hydrogeochemical studies in the region (Inoubli et al. 2006; Gouasmia et al. 2006) indicate that the regional aquifer, likely supplying the water required for carbonate precipitation, is hosted in the reefal, fractured, and karstified Aptian limestones, that host Na–Cl and Na–(Ca)–Cl–(HCO3−) waters at normal subsurface temperatures (29–39 °C). These waters have a high pCO2, facilitating carbonate dissolution at depth (Inoubli et al. 2006), and degas, and evaporate on emergence, thus facilitating the precipitation of carbonate, and some positive covariation of δ13C and δ18O as observed for samples labeled “Travertine” in Fig. 6. The presence of travertine textures (cf Folk et al. 1985) seems to support some deep CO2 contribution to the basin. Travertine deposits are associated with volcanism, deep faulting, metamorphism, and, often, with hydrothermalism, and tend to be enriched in 13C (− 2 to + 10‰, showing a median of + 4.3‰), which is explained by water–rock interaction and decarbonation reactions (Pentecost 1995, 2005, and references therein; Özkul et al. 2014). However, the interpretation of δ13C values > 4‰ remains somewhat speculative, because these higher values are found both in highly productive sub-environments (microbialite facies) and calcrete–palustrine facies. This fact would indicate that waters from which BEU carbonates precipitated had a relatively homogeneous composition that was differentially modified along the environmental spectra in the basin.

Oxygen isotopes

A reasonable estimate of the δ18O value of HCO3− available for the different carbonates studied might be derived from that of the Cretaceous limestones in the area of BEU. While discussing δ13C values above, it was argued that sample CRT 3 might well record primary sedimentary values, while the carbon and oxygen values of the remaining Cretaceous limestones and those of the alluvial conglomerate were modified by interaction with freshwater. If such, it would be expected that the limestone with the lowest δ18O values (CRT 0; δ18O = − 6.8‰; Table 1) represents maximum exchange, and best approximates the isotopic nature of meteoric water in the area at the time of formation of BEU. Using the equilibrium equation of O’Neil et al. (1969), the δ18OSMOW value of water in equilibrium with CRT 0 should have been something between − 8.0 and − 3.5‰, for temperatures varying from 10 to 30 °C. Current aquifer temperatures have been reported between 20 and 40 °C (Crombie et al. 1997; Inoubli et al. 2006), with deeper waters having higher temperatures. Considering that spring water emerges at about 20 °C, a calculated δ18OH2O ≈ − 5.6‰ SMOW would result, which is slightly higher than that of current average precipitation in the region (δ18OSMOW = − 6.6‰; Hamed 2014), but similar to regional groundwater (δ18OSMOW ≈ − 6.0‰; Ayadi et al. 2018).

Current values of precipitation (IAEA’s Carthago Station) average δ18O = − 3.6‰, but can be as low as − 8.5‰, and karstic groundwater in the nearby Teboursouk area yielded an average δ18O of − 6.1 ± 0.5‰ (Ayadi et al. 2018), consistent with oxygen values measured on the weathered Cretaceous basement. The present semiarid conditions of the area, however, are not thought to be comparable with those prevalent at the time of BEU formation. The Pleistocene–Holocene humid periods mentioned before imply very different climatic models, with humidity brought to the area largely by westerly winds of Atlantic provenance, rather than the current situation, where Mediterranean humidity, delivered by winds with a northerly trajectory, is the norm. The rain-out of humidity in an Atlantic–westerly scenario results in precipitation with δD and δ18O that are lower than in modern precipitation. The data of Sonntag et al. (1978) would predict δ18O values about − 8/− 9.5‰ at the longitude of Tunisia. Allowing for some Mediterranean contribution to groundwater (Crombie et al. 1997; Celle-Jeanton et al. 2001), with a weighted mean of δ18O ≈ − 6.6‰ (Hamed 2014), δ18O values of groundwater involved in the precipitation of BEU carbonates might well have had δ18O values around − 7/− 8‰.

The mineralogical characteristics of the facies that are formed by low-magnesium calcite (and no dolomite) suggest open fluvial-lacustrine environments with a minor effect or local effect by evaporation. Only salt lakes show marked fluctuations in the isotopic composition by long-term changes in the evaporation/precipitation ratio (Li and Ku 1997; Leng et al. 2005). In contrast, lakes with a significant throughflow show minimal variations attributable to changes in temperature or δ18O in precipitation, since lake water in open systems is generally similar to mean weighted annual precipitation.

The similarity of δ18O values of Cretaceous limestones with those of the carbonate facies of the Borj Edouane Unit (BEU) suggests that the Cretaceous carbonates of Gara Basin watershed were affected by a telodiagenetic process influenced water of weathering history similar to that of water that produced the carbonate facies of BEU. The lesser modification in δ13C with respect to the presumed original isotopic values may be due to the low C content, compared with the high O content, of the weathering fluid that affected the Cretaceous carbonate rocks. The correlation of relatively high δ18O values in some samples (TJCB3: − 2.58‰; Holocene TUFF: − 2.80) with high values of δ13C (11.68 and 12.46‰, respectively) is likely associated with episodes of greater evaporation and/or higher degassing rates. To this end, several authors agree that Pleistocene–Holocene humid periods in the North of Africa, as recorded by groundwater carbonates, were less intense, and the carbonate record less extensive, the younger they are (Szabo et al. 1995). Pluvial intensity decreased with time, with water in the youngest lakes having the shortest residence time, that resulted in increases in δ18O (Geyh and Thiedig 2008), such that reconstructed levels in some North African lakes may be better proxies for evaporation than they are for humidity (Dearing et al. 2001). If we keep in mind that travertines, due to evaporation of the water and supersaturation with CO2, frequently precipitate out of isotopic equilibrium (Chafetz and Lawrence 1994; Crombie et al. 1997), the high δ18O values of some of the samples measured (Table 1; Fig. 6) are not unusual.

Paleoclimate implications and age constraints

Unfortunately, the chronostratigraphic age is not well constrained for this unit and, at present, only an uncertain placement within the Middle–Late Pleistocene can be provided (Burollet and Sainfeled 1956; Ben Haj Ali et al. 1985). The fluvio-lacustrine–travertine lithofacies ensemble preserved in the BEU allows the interpretation that more humid conditions than presently prevail dominated in the north hinterland of Tunisia during its deposition. The fluvio-lacustrine microbialite deposits (oncoids and planar stromatolites) cropping out in the central area of BEU (central basin) show no evidence of exposure features, and their formation cannot be explained by a dry climatic scenario with water shortage. In fact, microbial fluvio-lacustrine structures are characteristics of relatively stable lakes “when precise conditions are met” in them (Casanova and Hillaire-Marcel 1992a), provided that enough Ca2+ is available to maintain saturation with respect to calcite (Casanova and Hillaire-Marcel 1992b). These microbial structures prosper within the photic zone of lakes with calcium-rich waters and reduced influx of siliciclastics, as in situ growths encrusting different substrates (Dean and Fouch 1983; Gierlowski-Kordesch 2010). The associated travertine facies at the eastern area of the basin were linked to the groundwater emergence from spring-fed sheet streams that flowed westerly (central basin). These springs were related to the fault network of the southeastern zone of the basin (Figs. 2, 7). The permanent water emergence from springs in combination with enough residence time of rainwater in the aquifer to become rich in calcium is necessary conditions for the travertine construction (Jones and Renaut 2010; Nicoll and Sallam 2017). Only under wetter conditions could the springs feed the area to the extent of sustaining a stable lacustrine zone in the central basin. From the environmental characteristics of this depositional system, and assuming a Middle–Late Pleistocene age, BEU should have accumulated during some of the Saharan humid periods, that have been attributed to a strengthening of the African Atlantic monsoon related to orbital increases in summer insolation (deMenocal and Baker 2000) in North Africa during Pleistocene time (see Grant et al. 2017 for a review, and references therein). In essence, the North African region was subjected during Pliocene and Pleistocene time to dry–wet oscillations regulated by orbital variations. Boreal summer rainfall stands out in this context due to the seasonal migration of the intertropical convergence (deMenocal 2004). In central Tunisia, abundant fossil lacustrine deposits around 90–100 ka and 130–150 ka have been recognized (Causse et al. 2003). In the studied zone, the Sarrat Wadi, flowing toward the Mediterranean Sea, has dissected the BEU and its Cretaceous basement to a depth of about 50 m below the top of BEU. Some 23 km upstream, the wadi cuts an endorheic lacustrine formation that hosts the paleontological and archeological site of Wadi Sarrat, dated to the beginning of the Middle Pleistocene (~ 0.7 Ma) by paleomagnetism and biochronology (Martínez-Navarro et al. 2014). However, the fossiliferous features and scant lithological description (“palustrine lacustrine dark unit”) of this lacustrine perennial facies are very different from those described at BEU. The fossil assemblage at Wadi Sarrat represents a period of more humid and temperate climate than today, likely related to an interglacial period (Martínez-Navarro et al. 2014). None of the species represented in this site are present in the fluvio-lacustrine microbialite facies of the BEU; thus, a different climatic context and, therefore, age of formation for both nearby deposits may be suggested. The travertine deposits of central Tunisia are linked to hydrothermal fluids associated with fault fissures, and are the first known and sedimentologically documented travertine facies in this region; thus, these deposits represent evidence of a humid episode during Early-to-Middle Pleistocene time (Henchiri et al. 2017).

Outside Tunisia and within the North African region, lacustrine and tufa facies are present at various chronostratigraphic levels within the Middle–Late Pleistocene, which have been commonly assigned to humid periods. In South Libya, the four members of a carbonate lacustrine succession have been dated at 420 ka, 380–290 ka, 260–205 ka, and 140–125 ka, correlated with 100-ka eccentricity cycles during the last 500 ka (Geyh and Thiedig 2008). In southwestern Egypt, some outcrops of tufa indicate pluvial episodes during Middle Pleistocene-to-Holocene time. In NW Sudan-SW Egypt, five lacustrine events occurred at about 320–250, 240–190, 155–120, 90–65, and 10–5 ka., four of which may be correlated with major interglacial stages 9, 7, 5e, and 1 (Szabo et al. 1995). In the nearby Kurkur Oasis, Crombie et al. (1997) recognize three spring-tufa units dated at 70–160 ka, 191–220 ka, and > 260 ka that they correlated with interglacial periods; these tufas are massive, containing vegetation-rich phytoherms, and stromatolite structures (Nicoll and Sallam 2017). About 230 km northwest of Kurkur, some fossil-spring tufas crop out in the Kharga Oasis that are correlated with episodes dated around 125 ka in relation with the marine isotope stage (MIS) 5e, although some spring deposits were formed during the MIS 6/5e transition, previous to the warmest last interglacial period (Smith et al. 2004, 2007). During this time, the south region of Egypt received enough precipitation to support perennial springs. In the tropical lake Manyara (Tanzania), generations of stromatolites have been assigned to two humid episodes dated ca. 90 and 140 ka, respectively, that were likely linked to peaks of the Indian summer monsoon (Casanova and Hillaire-Marcel 1992a).

The above-mentioned studies document that during the Middle–Late Pleistocene age proposed for the BEU (Burollet and Sainfeled 1956; Ben Haj Ali et al. 1985), there were various humid episodes represented by lacustrine and tufa carbonate deposits in the central and eastern North Africa, especially at the Middle–Late Pleistocene transition. Three of them are amply recognized between 380 and 120 ka (Geyh and Thiedig 2008; Crombie et al. (1997), which is why it is proposed that BEU was developed during one, or more, of these humid phases. However, proper age placement requires additional chronological studies that are being developed.

Conclusions

The Middle–Late Pleistocene BEU is the first continental carbonate system in North Tunisia to be documented, and constitutes a depositional system that shows an asymmetrical distribution of facies from west to east in the basin, arranged along four N–S traverses. The calcrete–palustrine association extends along the western traverse, while the microbialite association shows up in the central and eastern areas of the basin. The stromatolites (oncoids and planar stromatolites) occupy the central north–south traverse of the basin, changing to the easternmost travertine traverse through a transitional travertine–stromatolite zone.

Along a W-to-E transect, the depositional system changes from shallow intermittent lakes (calcrete–palustrine facies) through shallow permanent lakes, dominated by oncolitic structures, to the spring-tied travertine constructions in the eastern area. The distribution of environments reflects a trend in the diminution of water availability from east to west in the basin, which was due to spring groundwaters emergent along the fault network.

The δ18O values are restricted to a narrow range—around − 5.8‰—for all facies associations, except for a single value in a Cretaceous limestone, that is interpreted as the most primary-like value. Lower values in the host limestones are consistent with a meteoric overprint that seems to reflect the signature of paleorainwater sourced in the Atlantic Ocean, modified toward slightly higher oxygen isotopic ratios by evaporation and/or mixing with some additional recharge by waters of northerly (Mediterranean) provenance. The faint trend from this reference value toward higher δ18O in BEU carbonates reflects evaporation in the palustrine–lacustrine–travertine environments. In contrast, the δ13C values show a wide spread (− 6.4 to 11.7‰) for all facies, and there is an overlap between the lower values of the calcrete–palustrine association and the maximum values corresponding to the microbialite association. It is worth noting the relatively high carbon-isotope ratios for all facies associations, which are attributed to the predominance of carbon derived through decarbonation of carbonate bedrock of the Cretaceous substrate. In the calcrete–palustrine environment, the values were lowered by the contribution of additional CO2 derived largely from C4 plants, while in the microbialite associations, the higher δ13C carbon values were due to the combined degassing by evaporation and photosynthesis, especially in the central lacustrine areas.

From the analysis of sedimentological features and stable isotopes presented here, a more humid paleoclimate than in current times can be inferred for the deposition of BEU during Middle–Late Pleistocene time, when the area was influenced by westerly air flows sourced in the Atlantic Ocean.