Influence of biotic vs abiotic processes on the genesis of non‐marine carbonates along the Cameroon Volcanic Line (Cameroon) and palaeofluid provenance

Continental spring carbonates are perfect examples of the interaction of biotic and abiotic processes, and they preserve evidence of the velocity of the flow and the chemical composition of the spring water. This study focuses on non‐marine carbonates from fossil and active springs from the Bongongo and Ngol areas along the Cameroon Volcanic Line in South‐West Cameroon. Here, hydrothermal fluids reach the surface giving rise to small thermal springs, with temperatures between 31 and 49°C, and streams creating waterfalls, terracettes and barrage carbonate deposits. Petrographic analyses of these carbonates revealed that they are made up of stacked laminae of fibrous coarse crystals of low‐Mg calcite and laminae of alternate microsparite and micrite. The fibrous coarsely crystalline calcite, often with feather‐like fabric, grows from thin layers of micrite and peloids. Filaments of putative microbial origin are preserved within this peloidal micrite. The laminated microsparite and micrite microfacies is characterised by an intricate mesh of hollow filaments of microbial origin. The long feather‐like crystals of calcite formed in fast‐flowing water where the enhanced CO2 degassing has favoured the precipitation of CaCO3. The laminated micrite and microsparite, on the other hand, are probably formed in ponds where degassing and CO2 removal was lower and the calcite precipitation was fostered by microbial activity. The fast‐forming carbonates show higher Ce contents and very low total rare earth elements, revealing a preferential uptake of Ce with respect to other rare earth elements. This process would explain the positive or null Ce anomaly in continental spring carbonates elsewhere. The geochemical composition of these carbonates can be used as proxy for the characterisation of fluid/rock interactions between the groundwater and the substratum and for the characterisation of the sources of calcium and other elements that constitute tufa and travertines. The samples from Ngol are characterised by light rare earth element enrichment while those from Bongongo are overall enriched in heavy rare earth elements. Carbonates from both localities have a strong positive Eu anomaly (>4), suggesting a contribution from deep‐seated, hydrothermal, crustal fluids in contact with volcanic rocks and the breakdown of plagioclase from the Cameroon Volcanic Line alkali basalts.


| INTRODUCTION
Terrestrial carbonates (i.e. carbonates that form in continental settings) and particularly spring carbonates are deemed to be good geological repositories of climate and environmental data (Brasier, 2011;Camuera et al., 2014;Capezzuoli et al., 2010Capezzuoli et al., , 2014Gruszczynski et al., 2004;Smith et al., 2004aSmith et al., , 2004bWright & Tucker, 1991). Continental carbonates are also useful proxies for the characterisation of fluid/rock interactions between the groundwater, formation water and the rock substratum (Kokh et al., 2017;Teboul et al., 2016;Uysal et al., 2007Uysal et al., , 2009) and for the characterisation of the sources of major and trace elements that constitute the spring deposits (Brogi et al., 2016(Brogi et al., , 2020Teboul et al., 2016). Continental carbonates have the potential to illustrate the evolution of non-marine organisms from the Archean to present as well as provide evidence for non-marine life during periods of intense climate and environmental stress (Capezzuoli et al., 2010;Smith et al., 2004aSmith et al., , 2004b and valuable information about non-classical crystallization pathways with important ramifications for astrobiological research (see discussion in Franchi & Frisia, 2020).
It is now clear that, in order to use carbonates as reliable geochemical proxies, it is crucial to distinguish between inorganic vs biomediated mechanisms of mineral precipitation (Chafetz & Guidry, 1999;Pentecost, 1990;Shiraishi et al., 2019). The accuracy of spring carbonates as archives of palaeoenvironmental information depends on their mode of formation and the interplay of abiotic and biotic processes (Brasier, 2011;Della Porta, 2015 for a comprehensive review). The mineralogical composition, as well as the facies and microfacies assemblages of spring deposits, is linked with variations in environmental parameters such as composition and temperature of spring water, CO 2 degassing, calcite saturation index (SI), macrophyte activity and microbial metabolism (Camuera et al., 2014;Capezzuoli et al., 2010Capezzuoli et al., , 2014Jones & Renaut, 2010;Turner & Jones, 2005). The formation of carbonates in continental springs may be driven exclusively by extrinsic processes, and therefore characterised by inorganic precipitation of carbonates, or follow the processes of organomineralization sensu lato that include biologically induced and biologically influenced mineralization, depending on the role played by living organic matter (Dupraz et al., 2009). The role of organic compounds, either of biological or inorganic origin, in the crystallization of carbonates in continental environments is still debated, although recent breakthroughs due to the application of nano-scale techniques revealed the importance of non-classical crystallization pathways (De Yoreo, 2013;Rodriguez-Navarro et al., 2016). This non-classical nucleation theory suggests that calcium carbonate crystallization is driven by degassing and by changes in pH and calcite SI (Wolf et al., 2008), without the active participation of any organic compound. On the other hand, the organic compounds, and especially microbial exopolymeric substances (EPS) are proven to act as catalysers for the development of fabrics such as laminites, often considered as microbialites (Dupraz et al., 2009;Riding, 2011;Turner & Jones, 2005).
In this work, petrography and geochemistry of spring carbonates from the Cameroon Volcanic Line (CVL) from the Ngol and Bongongo areas ( Figure 1) are investigated to identify the processes that induced their precipitation and the provenance of the parent fluids. These carbonates formed from springs of hydrothermal water with temperature above 30°C and are, therefore, considered as travertines (Capezzuoli et al., 2014;Pentecost, 1990;Shiraishi et al., 2019). The occurrence and bulk composition of travertines from Bongongo and Ngol were presented for the first time by Le Maréchal (1971, 1976 and more recently by Bisse et al. (2018). Bisse et al. did not provide details on the petrography and facies analyses of the carbonates and the geochemistry and petrography were not linked with their genesis and fluid provenance. Here, novel petrographic observation and rare earth elements (REE) geochemistry of the CVL carbonates are coupled with literature data in an effort to address a knowledge gap concerning the variability of the carbonate depositional environments along the CVL and the genesis of the spring water. waters which are characterised by relatively low depositional rates producing highly porous bodies with poor bedding and lenticular profiles…" (Capezzuoli et al., 2014, p. 3).
The term 'mound' in continental settings (Crombie et al., 1997;Keppel et al., 2012;Pentecost & Viles, 1994 and references therein) refers to mounded morphologies as consisting of continental carbonates associated with either ambient temperature groundwater or hydrothermal water discharges.
The term micrite is used here with reference to calcite crystals with a diameter of 4 μm or less (microcrystalline calcite). The term microsparite is used here to identify a calcite matrix (as seen under the optical microscope) with uniform crystal size (between 4 and 62 µm) and equant crystal shape (Dunham, 1962;Flügel, 2004).
The term peloid (not including pellets) is used here to identify grains composed of cryptocrystalline and microcrystalline carbonates regardless of origin (Bathurst, 1971). Clotted peloidal micrite is a textural term that refers to a fabric consisting of peloids, typically 20-60 µm in size (Riding, 2000), forming amalgamated clots (Flügel, 2004).
The term phytoherm is used following the definitions in Ford and Pedley (1996) and Keppel et al. (2011) to identify a laminated carbonate characterised by calcite-coated roots and stems of hydrophytes that were bound when the carbonates were deposited, hence the use of 'boundstone' (cf. Franchi & Frisia, 2020). In this facies, the vegetation living in pools and streams offers a favourable substrate for the growth of spring carbonates. F I G U R E 1 Geographic location and schematic geological map of the study area. (A) Location of Cameroon (outline) in western Africa. (B) Location of the study area (C) along the Cameroon Volcanic Line (grey areas). (C) Schematic geological map of the studied area. Modified from Le Maréchal (1976) and Bisse et al. (2018) | 105 BISSE Et al.

| Geological setting
The study area is located in south Cameroon in the southern, continental part of the CVL ( Figure 1A,B). The CVL is a Y-shaped volcanic/tectonic megastructure (Moreau et al., 1987) oriented N30E. It is formed by a chain of oceanic islands (Pagalu, Sao-tome, Principe and Bioko) which date from Upper Eocene to Recent (Déruelle et al., 1991) (Lee et al., 1994). The CVL extends from Mount Cameroun in the south-west to Lake Chad in the north (Dorbath et al., 1986). Although there is no consensus on the real size of the CVL previous works have estimated lengths of between ca 1,200 and 2,000 km (Fitton, 1980;Moreau et al., 1987;Tchoua, 1976). In the study area ( Figure 1C), the Precambrian granitoid basement is unconformably overlain by Lower Cretaceous to Neogene sediments and by Cenozoic volcanic rocks of the CVL (Tchoua, 1976). The basement granites are composed of K-feldspars, plagioclase, quartz biotite and hornblende (Tchameni et al., 2001). The dominant lithologies among the CVL rocks are syenite and basalt while the sedimentary cover is mainly made up of shales, sandstones and alluvium ( Figure 1C) (Bisse et al., 2018;Le Maréchal, 1976). The mineralogical composition of the basalts is dominated by olivine, augite and plagioclase. The syenite is made up of sanidine, quartz, clinopyroxene and aegerine (Kamgang et al., 2010(Kamgang et al., , 2013Njonfang et al., 2013). A detailed description of the mineralogy of the volcanic rocks exposed in the area is provided by Kamgang et al. (2010Kamgang et al. ( , 2013 and Njonfang et al. (2013).
The Palaeogene-Neogene and Lower Cretaceous sandstones ( Figure 1C) are mostly sub-arkosic sandstones dominated by feldspars, quartz and micas (biotites and muscovites) with ferruginous and carbonate cements (Ngueutchoua et al., 2017). The Upper Cretaceous shales are mostly made up of quartz, feldspars and calcites. The alluvium (Palaeogene to Neogene) is mostly quartz-rich sand with minor amounts of feldspars, muscovite and biotite (Ngueutchoua et al., 2017).

| Ngol spring
At Ngol, an active spring with a diameter of ca 1 m ( Figure 2A) feeds a system of small streams and ponds. Samples in this locality have been collected from both active streams, relict (fossil) streams and travertine mounds ( Figure 2B through F). The drainage system was modified by mining activities during the Second World War (Bisse et al., 2018). The water from the Ngol spring has an average temperature of 31°C and a pH of 6.2 measured by Bisse et al. (2018) at the spring orifice ( Figure 2A). Samples of laminated travertines have been collected from small cascades and terracettes ( Figure 2D) and along the main stream where carbonates encrust the country rock and the debris along the banks of the stream ( Figure 2E). Some of the carbonate deposits along the bank of the river appear as poorly lithified, massive to faintly laminated, porous carbonates covered in moss ( Figure 2F). At Ngol, the main lithofacies consist of phytoherm boundstone and laminated carbonates often engulfing plant remains ( Figure 2G,H). One of the samples selected for detailed petrographic analyses was taken from horizontal to sub-horizontal laminated carbonates from an active pool ( Figure 2G). Another sample was collected from the concretions found along an active cascade ( Figure 2B,H).

| Bongongo spring
The spring system at Bongongo is fed by an active, circular spring (ca 20 cm across, Figure 3A) with an average water temperature of 49°C and pH 6.2 measured by Bisse et al. (2018) at the spring orifice. The area is characterised by highly vegetated pools ( Figure 3B) and fast-flowing streams with cascades with active precipitation of carbonates ( Figure 3C). Fossil travertine mounds consisting of laminated travertine and phytoherm boundstone ( Figure 3D,E) and massive to faintly laminated carbonates ( Figure 3F) have been sampled. In this locality, finely laminated travertines are engulfing the plants remains are very common; they are characterised by alternating dark toned laminae and reddish laminae ( Figure 3G,H). Samples from this locality have been collected from a fossil cascade ( Figure 3G) and from the lower lithofacies of a fossil travertine mound ( Figure 3E,H).

| Water chemistry and CVL travertines isotopic composition: a background
In the first half of the past century and then again in the 1970s, extensive exploration was carried out in the thermal springs of western Cameroon with the intent of creating sanatoriums and exploiting the abundant thermal waters (Le Maréchal, 1971Maréchal, , 1976 and references therein). The geochemical composition of the carbonates from Ngol (Figure 2), located at ca 840 m a.s.l., and Lobé (Bongongo in this paper, Figure 3), located at 60 m a.s.l., is here reported in Table 1 (Le Maréchal, 1971Maréchal, , 1976 and references therein).
The active spring at Ngol was characterised by degassing (bubbling) and by precipitation of Fe-bearing colloids (Le Maréchal, 1971). The author describes heavily faulted crystalline bedrock (mainly granite, gneiss and migmatite) with active and vestigial springs aligned along the fault lines. The water from the Ngol spring orifice shows high Ca and Mg concentrations, low Na and very low Cl concentrations ( Table 1).
The early reports account for several vestigial springs in the Bongongo area and one active spring (see Figure 3A), with a discharge (at the date of the report, 2 April 1971) of ca 200 L/h, found on top of a conical build-up of spring carbonates (in the vicinity of the mound in Figure 3D; subsamples BSB1-BSB5 in Table 2). The water from the Bongongo spring orifice shows very high Cl and Na concentrations and moderate Ca and Mg concentrations (Table 1).
In more recent years, Fantong et al. (2019) reported the overall chemical composition of the water flowing from the active springs occurring along the CVL as falling into the evaporated Na+K-Cl and non-evaporated Ca+Mg-HCO 3 categories. Although the authors do not present data from Ngol springs, these recent analyses partially confirm the data presented in Table 1.

BISSE Et al.
The carbon and oxygen stable isotope composition of samples of carbonates from the Ngol and Bongongo springs determined by Bisse et al. (2018) are reported in Table 2. Although their research did not provide for detailed petrology of the carbonates, Bisse and co-authors noted that the stable isotopic composition of C and O from precipitates found at Ngol (δ 13 C between 0.4 and 1.2‰ PDB; δ 18 O between −8.3 and −5.8‰ PDB) and Bongongo (δ 13 C between 1.1 and 2.0‰ PDB; δ 18 O between −8.4 and −6.4‰ PDB) was similar, suggesting a common source of C for the two spring sites. It is important to note that the C and O stable isotopes value presented in Bisse et al. (2018) do not always overlap with the samples selected for petrographic analyses in the present work. Therefore, they cannot be linked to local degassing and precipitation processes but can shed light on the nature of the spring water along the CVL. Bisse et al. (2018) compared the isotopic composition of the travertines from Ngol and Bongongo springs with published data of travertine worldwide. The overall positive δ 13 C values from the CVL travertines suggest a thermogene spring nature for these carbonates (cf. Jones & Renaut, 2010;Kele et al., 2011;Pentecost, 1995;Pentecost & Viles, 1994).

| MATERIALS AND METHODS
Four field campaigns took place in the study areas of Bongongo and Ngol ( Figure 1C) between January 2013 and December 2016, mainly during the dry season (December to March).  Table 2 for sample details Samples of carbonates were collected from 10 localities near Bongongo and Ngol springs (Table 2). During the field studies, 18 samples were retrieved (Table 2), which were then micro-drilled to obtain subsamples for geochemical analyses.
Four specimens of well-lithified carbonates, two from Ngol ( Figure 2G,H) and two from Bongongo ( Figure 3G,H), have been petrographically characterised. Petrographic description of the spring carbonates follows the classification in Della Porta (2015), based on the classifications of Dunham (1962), Embry and Klovan (1971) and Wright (1992). X-ray diffraction (XRD) analyses (Table 2) aimed at characterising the mineral phases present in the specimens were performed with a Bruker D8 Advance X-ray diffractometer (Cu Kα X-ray source) at the Botswana International University of Sciences and Technology (BIUST) laboratories in Palapye. The range investigated was between 5° and 70° 2Θ with a step size of 0.03°. The raw XRD data were then processed by Rietveld refinement using the FullProf4 software (Rodríguez-Carvajal, 2001). The MgCO 3 content in calcite phases was obtained from the shift of calcite 112 peaks in the diffractograms (cf. Gayathri et al., 2007;Goldsmith et al., 1955;Taviani et al., 2015).
Geochemical data of major and trace elements from Bisse et al. (2018) were integrated with novel REE analyses (Table 3). These geochemical analyses were carried out at the Laboratory of Climate and Environmental Sciences (Direction des Sciences de la Matiere, Paris, France), using an inductively coupled plasma-mass spectrometer (ICP-MS). A 0.5 mg aliquot of rock powder was digested in 10 ml of nitric acid (in closed beakers). The instrumental precision of almost all elements was 5% for either five or six compiled solutions where the elements were above the limit of quantification. Geochemical data for the trace elements and REE are presented in Table 3. Trace and REE concentrations were normalised to Post-Archean Australian Shale (PAAS; Taylor & McLennan, 1985) and to Cl-chondrite (Anders & Grevesse, 1989) for comparison, as PAAS-normalisation can be problematic for the discussion of REE anomalies in carbonates affected by the circulation of hydrothermal fluids (see discussion in Tisane et al., 2019). Normalised La, Ce, Eu and Gd anomalies were calculated using the geometric equation given by Lawrence et al. (2006) (Table 3). The light REE (LREE) fractionation was calculated as Pr SN /Yb SN to avoid bias due to anomalous La and Ce concentrations (Franchi, 2018) and the ratio between medium REE (MREE) and heavy REE (HREE) was calculated as Gd SN /Yb SN (Table 3).
The mineral chemistry of travertines from Ngol and Bongongo was analysed with a JEOL JXA-8230 electron probe micro-analyser (EPMA) at BIUST. The analytical conditions for EPMA analyses were as follows: 15 kV accelerating voltage, 2 × 10 −8 Å probe current, 10 ms dwell time, three accumulations. The analyses were carried out using spectra calibrated with a Serial IU-MINM25-53 (Astimex Standard Limited) mineral standard mount.
Selected samples of travertines from both Ngol and Bongongo were imaged with a scanning electron microscope (SEM) at the Microscopy and Microanalyses Unit at the University of Witwatersrand (South Africa). The samples were mounted on lipped 32 mm aluminium stubs using one or two disks of 12 mm carbon tape (SPI Supplies), and the top of the samples bridged to the stub with 12 mm carbon tape, cut to 6 mm wide (TAAB). The samples were coated with 5 nm carbon in an EmiTech K950X Turboevaporator  Gazel (1947), Dumort (1965) and Le Maréchal (1971 (Quorum Technologies), followed by 5 nm 60/40 gold/palladium in an EmiTech K550X Sputtercoater. Samples were mounted at a working distance of 10 mm in a FEI Quanta ESEM, and imaged at 30 and 20 kV. All imaging was performed using backscatter detection at a spot size of 4.0 or 4.5.

CVL carbonates
The carbonates collected from the Ngol and Bongongo springs and prepared for petrographic analyses appear as botryoidal to laminated concretions consisting of stacked millimetre-thick laminae of carbonate cement and micrite ( Figures 2G,H and 3G,H). The samples analysed from the Ngol area are made entirely of low-Mg calcite with a composition close to pure calcite (MgCO 3 contents lower than 2%; Table 2). The carbonates collected from the Bongongo area are made up of low-Mg calcite but show higher MgCO 3 contents with respect to Ngol samples, with values up to 4% (Table 2). Plant stems were found radially coated by carbonates leaving a mouldic porosity after their decay ( Figures 2G,H and 4A). These travertines consist of up to 1 cm thick stacked laminae constituted by two distinct types of microfabrics: (a) fibrous coarsely crystalline calcite with elongated crystals (feather-like fabric) (Figures 4 through 6) and (b) laminated filamentous-rich micrite and microsparite (Figures 7 through 10).

| Fibrous coarsely crystalline microfacies
This microfacies is made up of low-Mg calcite, although Mg concentrations appear to rise along the edges of the fibrous, elongated crystals of calcite ( Figure 5).
The lamination in the fibrous coarsely crystalline microfacies is given by clear calcite alternating with turbid micrite laminae ( Figure 4B through D). The laminae engulfing the plant remains are made up of dark brown, microcrystalline calcite and are overgrown by crystal fan calcite (<500 µm long, Figure 4A) and by fibrous calcite with elongated crystals (>500 µm long) forming peculiar feather-like fabric and dendritic fabric (cf. Camuera et al., 2014) (Figure 4B,C). The feather-like crystals grew on thin micrite laminae or on plant tissues and have clear accretionary laminae marked by micro-inclusions and micrite films ( Figure 4B through E). The crystals are engulfing <10 µm-thick filaments longitudinally that can run for the entire length of the crystal ( Figure 4F,G). The fibrous calcite in Figure 4G  sample stage of the microscope is turned counterclockwise, a characteristic of fascicular-optic fibrous calcite sensu Richter et al. (2011). The fibrous calcite crystals from the Ngol carbonates have a length-width ratio >10:1 and show a typical bifurcation toward the end ( Figure 4B; split crystal or crystal splitting, cf. Kendall, 1985). The Bongongo counterparts are up to several millimetres long, and are better described as feather-like crystals with a length-width ratio >20:1 ( Figure 4G).
The stacked laminae of fibrous and feather-like calcite crystals alternate with laminae of micrite/microsparite and clotted peloidal micrite and some of these calcite crystals grew upon micrite peloids ( Figure 4C,H). The micrite peloidal fabric, which is interlayered with the large calcite crystals, preserves isolated armoured filaments (sensu Franchi & Frisia, 2020), coated by euhedral crystals or completely encrusted by micrite ( Figure 6B,C), and tangles of non-armoured filaments ( Figure 6A through C). Peloids are made of microcrystalline calcite and by a mesh of acicular crystals ( Figure 6D,E). The microsparite calcite crystal present micron-sized rounded holes that seems to be pervasive in this microfacies ( Figure 6F).
Plant cellular structures are preserved as moulds in the coarse calcite facies ( Figure 6G). These plant moulds are disseminated by microboring structures and partially filled by microcrystalline acicular cements ( Figure 6H).   (Figure 7). This laminated micrite microfacies shows little evidences of plant mouldic porosity when compared to the fibrous coarsely crystalline facies. The reddish micrite and microsparite are made by low-Mg calcite with abundant fine and very fine (<10 µm) grained clasts of terrigenous materials (Figure 7).
Reddish micrite and microsparite preserve filaments (a few microns thick and up to 100 µm long) that grow quasi perpendicular to the laminae ( Figure 7A through C). These filaments are pervasive and cut across surfaces between the laminae of microsparite and micrite ( Figure 7B,C). The filaments are either filled by microsparite and/or micrite (red arrows in Figure 8D,E) or appear to be hollow ( Figure 9A through C). The layering is clearly related to changes in the porosity of the micrite laminae (cf. Perri et al., 2012) given, in this case, by the presence of circular microborings and moulds of filaments that pierce the microsparite crystals ( Figure 9B,C). Few preserved filaments of putative microbial origin have been identified in the microsparite ( Figure 9E,F). Some of these filaments appear as fresh filaments ( Figure 9E,F) and are probably not linked with the processes discussed in this paper. The crystals of microsparite are euhedral to subhedral, often bored by circular holes and have a peculiar lamellae overgrowth ( Figure 9F).
The laminated micrite and microsparite are characterised by isolated fans of clear, laminated microsparite growing perpendicular to the lamination ( Figure 7D). These microsparite fans are made up of low-Mg calcite and are truncated by a corrosion surface enriched in Fe-minerals ( Figure 10). The lamination within the fans is marked by changes in Ca and Fe concentrations ( Figure 10D through F).

| Ngol springs
The carbonates from Ngol springs show Li contents varying between 0.75 and 2.14 ppm and Na contents varying between 21.55 and 65.36 ppm ( Table 3). The Mg contents vary between 837 and 1,850 ppm while Sr varies between 360 and 681 ppm (Table 3). Barium shows concentrations comprised between 22.76 and 54.06 ppm (Table 3). There is no significant correlation between Na and Mg, Sr and Mg, Ba and Mg, whereas Li shows a strong positive correlation with Mg (R 2 = 0.9; Figure 11A through D). The trace element contents normalised to the PAAS show an overall depletion with respect to the reference material, except for Sr where the concentration is strongly enriched ( Figure 12A).
The ∑REE content shows high variability going from as low as 0.83 to 10.81 ppm ( Table 3). Samples of travertine from the Ngol area show a consistent PAAS-normalised REE pattern with slight enrichment of LREE and MREE with respect to HREE ( Figure 12B; Table 3). The Y/Ho ratio from the Ngol travertines shows overall near-chondritic values with an average of 48 (Table 3). Overall the La anomaly shows positive values (between 0.87 and 1.49) ( Table 3) Table 3). Interestingly the Eu anomaly is always strongly  Table 3). The samples of coarsely crystalline fibrous calcite from Ngol show a distinct REE pattern when compared to samples dominated by laminated micrite and microsparite, with a Ce anomaly near 1 and a very low ∑REE content ( Figure 12D).

| Bongongo springs
Laminated carbonates from the Bongongo area show Li contents between 3.76 and 6.82 ppm (Table 3). When compared with samples from Ngol these travertines show very high Na and Mg contents with values as high as 1,116 and 8,166 ppm, respectively (Table 3). The travertines from Bongongo springs show elevated concentrations of Sr and Ba when compared with samples from Ngol, with values ranging between 2,847 and 7,749 ppm and between 315 and 864 ppm, respectively (Table 3). There is no significant correlation between Na and Mg, Sr and Mg, Ba and Mg, Li and Mg ( Figure 11A through D).
The ∑REE content is generally below 2 ppm (Table 3). Samples from Bongongo springs show highly variable PAAS-normalised REE patterns with variable LREE/HREE ratios (between 0.31 and 2.93) and highly variable MREE/ HREE ratios (between 0.64 and 6.3; Table 3). The Y/Ho ratio of the travertines from Bongongo varies from chondritic to super-chondritic values (38-133; Table 3). The La anomalies vary between 1.14 and 7.17, whereas Ce anomalies are always around 1.3 (Table 3). The Eu anomaly is always strongly positive in both PAAS-and CL-chondrite-normalised patterns with values as high as 6.6 (Table 3; Figure 12B,C).

| Fibrous coarsely crystalline microfacies
The fibrous coarsely crystalline microfacies in both Ngol and Bongongo were identified mainly in samples collected from active and fossil streams and cascades (Figures 2 and 3). This fibrous coarsely crystalline calcite has formed along the streams where the surface area of the substrate in contact with the spring water is large ( Figure 2C,D). The large surface area/water depth ratio along the streams favoured degassing and, consequently, increased pH and the SI of calcite (Franchi & Frisia, 2020;Keppel et al., 2011;Pentecost, 2005) promoting calcite precipitation around filaments ( Figure 4F through H). On the other hand, it has been demonstrated that local, small-scale variations in CO 2 outgassing may not be coupled with local variations in the calcite precipitation rate (Hammer et al., 2008(Hammer et al., , 2010.
Changes in pH alone within the spring water column may foster both nucleation and growth and the carbonate phase that precipitates. Unfortunately, the existing water chemistry data are from the spring orifice and one can only assume an increase of pH downstream due to continuous CO 2 degassing.

BISSE Et al.
The coarsely crystalline feather-like crystals identified in both Ngol and Bongongo carbonates have a striking resemblance to the 'feather crystals' of calcite described in Kele et al. (2011) from the thermal springs of Pamukkale. This fibrous calcite is analogous, amongst others, with the 'crystalline facies' described in the Gran Canaria Island by Camuera et al. (2014) and with the 'rays' described by Chafetz and Folk (1984) from travertine deposits in Italy and the USA. The authors consider this coarse crystalline, fibrous calcite to be product of rapid precipitation and growth from supersaturated water due to rapid CO 2 degassing favoured by fast-flow conditions (Camuera et al., 2014;Chafetz & Folk, 1984;Okumura et al., 2012). Under such high flow velocity conditions, the dominant processes are abiotic and lead to the formation of coarser fabrics (Okumura et al., 2011;Shiraishi et al., 2019). Richter et al. (2011) consider fibrous calcite to be formed by rapid crystal growth. Carbonate crystals formed under a fast precipitation rate incorporate increasing quantities of Mg into their lattice and Mg plays a significant role in the formation of fibrous fabrics (Richter et al., 2011). In the spring carbonates from CVL, the samples from active streams and cascades (e.g. BSN2 and BSN3, Table 2), and those with coarse crystalline fibrous calcite (e.g. BSN4), have higher Mg contents when compared to samples of laminated, filamentous-rich micrite and microsparite (Table 3). This excess in Mg might account for the formation of fibrous calcite (see discussion in Richter et al., 2011).
The fibrous coarse crystals of calcite described in the samples from Ngol and Bongongo can be compared with the calcite dendrites associated with spring deposits in the Kenyan Rift Valley, Iceland and China described in Jones and Peng (2012). In this case the authors have also attributed the elongated, coarse crystals of calcite to rapid CO 2 degassing of CO 2 -rich, highly supersaturated spring waters (see also Jones & Renaut, 1995, 2010Jones et al., 2005).
The fibrous coarse crystalline microfacies in both the Ngol and Bongongo deposits preserves evidence of microbial filaments (probably both fungal mycelium and cyanobacteria) trapped longitudinally within the crystals ( Figure 4F,G). The micrite-encrusted filaments identified in this microfacies are mainly found in the micrite interlayers and within the peloidal fabrics ( Figure 6A through C). These encrusted filaments show some analogies with the 'armoured' filaments found in the Great Artesian Basin tufa (Franchi & Frisia, 2020). Dupraz et al. (2009) suggested that the presence of encrusted, rather than impregnated, filaments in microbialites is related to fast-flowing water with fast CO 2 degassing. In the absence of direct measurements of stream water chemistry, the presence of encrusted filaments, reinforces the interpretation that coarse crystalline, fibrous calcite was indeed formed during periods of high discharge with fast-flowing water and rapid CO 2 degassing.
The carbonates studied from Bongongo springs have slightly different fabrics showing the alternation of featherlike crystals adopting dendritic (sensu Camuera et al., 2014) fabrics made up of vertical filaments overgrown by sparite ( Figure 4G). The site for calcium carbonate growth, in this case, is provided by organic filaments (Camuera et al., 2014;Pentecost, 2005) but the mechanism favouring the calcite growth is still the rapid CO 2 degassing (see also Turner & Jones, 2005). In fact, the strong elongation of the calcite crystals primary stems and the small side branches are indicative of strong competition for growth space among single crystals (Jones & Peng, 2012), hence the 'feather-like' appearance (cf. Camuera et al., 2014;Kele et al., 2011). The increase in the length-width ratio from Ngol to Bongongo might denote an increase in supersaturation of the spring water and fast CO 2 degassing from CO 2 -rich water (Jones & Peng, 2012;Jones & Renaut, 1995). The precipitation and growth of calcite along the active streams of Ngol and Bongongo is not necessarily driven by intrinsic metabolic processes, but rather by extrinsic factors (see discussion in Franchi & Frisia, 2020). On the other hand, the presence of filaments is thought to support the vertical aggradation of the carbonate fostering the growth into stacked laminae as it has been demonstrated that the daily migration cycle of cyanobacteria forms the lamination in travertines (Shiraishi et al., 2019 and references therein). The presence of microbial filaments (including fungal hyphae, cf. Riding & Awramik, 2000), and at a larger scale the roots and stems of macrophytes, can provide favourable sites for calcium carbonate growth into dendritic and laminated fabrics (Turner & Jones, 2005), but it is the change of SI C and pH associated with degassing of CO 2 , instead of microbial-related organic compounds (i.e. EPS) and metabolic processes, that promoted the growth of coarse calcite crystals (Franchi & Frisia, 2020;Gebauer et al., 2014;Wolf et al., 2008).
The alternation of phases where fast growing, vertical calcite crystals are interspersed with intervals of etching/corrosion and the formation of micrite and other carbonate phases such as in the CVL carbonates ( Figure 4B through D) has been directly related to variations in the composition of the spring hydrothermal water (Jones & Peng, 2012). The numerous laminae of micrite that interrupt the otherwise clear crystals of fibrous calcite ( Figure 4B through D) are evidence of rapidly changing water composition (i.e. Fe content) and might reflect annual growth cycles (Camuera et al., 2014;Jones & Renaut, 2008;Jones et al., 2005) or changes in the water discharge rate (Pentecost et al., 1997). These micrite laminations ( Figure 4B through D) formed at times when saturation levels in the spring water decreased and the growth of fibrous and feather-like calcite crystals ceased (Jones & Peng, 2012).
The micrite and microsparite from the peloidal fabric presents widespread, small (a few microns across) rounded holes ( Figure 6) that can be interpreted either as microboring (c.f. Camuera et al., 2014) or traces of diatom pedicles (cf. figs 4B and 5A in Pedley, 2000).

| Laminated filamentous-rich micrite microfacies
This microfacies has been identified in carbonates from active pools ( Figure 2G). The presence of abundant putative microbial filaments into the micrite laminated microfacies from a pool near Ngol springs ( Figures 7A through C and  9), suggests that the microbial communities (including fungal mycelium) might have exerted a major catalytic function on the growth of the calcite in this microfacies. This microfacies presents an intricate network of calcified, hollow filaments in both transverse and longitudinal sections (Figure 9). These filaments are completely engulfed in micrite and microsparite ( Figure 9B,C) but they also pierce crystals of sparite and microsparite ( Figure 9C through F). It appears, therefore, and fine (BSN5-9) calcite microfabrics from Ngol that the microbial mat directly influences the growth process responsible for the lamination of these travertines whereby, intervals of micrite crystals nucleating within the bacterial EPS are followed by periods of higher physical-chemical instability characterised by the growth of sparry calcite and the abiogenic growth of calcite on biogenic supports (i.e. filaments; cf. Camuera et al., 2014).
There is no doubt that filaments (either microbial or fungal in origin), and plant roots and stems, provide favourable sites for the nucleation and growth of calcite crystals when waters may be barely at saturation or even undersaturated with respect to calcium carbonate (Camuera et al., 2014;Gradziński, 2010;Pedley, 1992;Pentecost, 2005). The filaments in the micrite laminated facies show evidence of impregnation of bacterial sheaths ( Figure 9E). The impregnation of filaments can be facilitated in slow-flowing, CO 2 -poor streams or lakes (Dupraz et al., 2009). Under such conditions, the fractionation of REE into the carbonate phases would be substantially different from the fast-forming coarsely crystalline calcite (see below).
The laminated filamentous-rich micrite microfacies was formed in pools and ponds, where the flow velocity is limited, characterised by lower CO 2 degassing and consequently by lower saturation rates. As a result, travertines are often botryoidal and/or with porous texture, reflecting a diffusiondominated system (Hammer et al., 2010). This condition would favour microbially influenced precipitation of micritic fabrics (Capezzuoli et al., 2014;Chafetz & Guidry, 1999;Della Porta, 2015;Hammer et al., 2010;Shiraishi et al., 2019). The low variability in δ 13 C values seems to disagree with the hypothesis that different microfacies were formed under an increasing degassing rate. Higher degassing would in fact result in higher δ 13 C values (Kele et al., 2011 and references therein). Unfortunately, the available δ 13 C isotopic data do not correspond to the lithofacies described here, hence they cannot be used to support or disregard this hypothesis.
Looking at the arrangement of the filaments in the micrite laminated facies of the CVL travertines (Figures 7 and 9) it appears that microbes had to pierce the laminae to avoid getting engulfed in the calcite precipitate, a mechanism similar to the accretion of stromatolites (Reid et al., 2000) promoting the formation of stacked lamina. Otherwise, this pervasive network of filaments might be the result of endolithic communities (Pentecost et al., 1997). In the absence of wellpreserved filaments, it is impossible to proceed with a taxonomic identification of the microbial communities.
On the other hand, the lamination and the changes from laminated micrite to coarse crystalline microfacies might have been a result of entirely inorganic processes linked to frequent changes in the chemistry of the spring water (Guo & Riding, 1992;Pentecost et al., 1997). The presence of brownish-black crusts of Fe-rich minerals above the microsparite fans (Figures 7 and 10), for instance, indicates stages of interrupted crystal growth or corrosion, due to variations in water geochemistry (Camuera et al., 2014), for example a drop in SI C and/or pH, lower flow rates. Shiraishi et al. (2019) have demonstrated that under lower flow velocities a micritic fabric with abundant interlaminar porosity forms under microbial influence. Under higher flow velocities, sparitic fabrics with little inter-crystalline porosity (more compact fabrics, cf. Hammer et al., 2010) formed through abiotic precipitation become dominant. The microbial processes that lead to the deposition of micritic fabrics are suppressed as the flow velocity increases giving way to abiotic processes and the consequent formation of coarse crystalline fabrics (Shiraishi et al., 2019). Such a notion goes against the finding of Pedley and Rogerson (2010) who proved that increasing water velocity can affect the fabric of precipitates that are, nonetheless, biologically influenced.

CVL carbonates
The geochemistry of carbonates, and in particular, the distribution of REE can be a powerful proxy for the investigation of the processes of formation and growth of spring deposits and it has been used in the past to assess the fluid/rock interaction between groundwater and the substratum (Franchi & Frisia, 2020;Kokh et al., 2017;Teboul et al., 2016;Uysal et al., 2007Uysal et al., , 2009. With the exclusion of speleothems, very few studies report the REE distribution of continental carbonates (Chagas et al., 2016;Franchi & Frisia, 2020;Uysal et al., 2007). Nevertheless, it is known from marine carbonates that the REE distribution can indicate a probable source of the parent water and highlight partitioning effects into the organic colloidal fraction and therefore shed light on the initial stages of crystallization and early diagenetic phases (Franchi et al., , 2016Hu et al., 2014;Kim et al., 2012;Pourret et al., 2008).
In organic-rich alkaline waters, the inorganic phase is always enriched in HREE; this results in a mirrored pattern in the organic colloids that will show an LREE enrichment (Pourret et al., 2008). At high carbonate alkalinity and a pH over 8.0, Ce(III) is oxidized to Ce(IV) that can be easily absorbed by organic matter leading to a positive Ce anomaly of the organic-rich colloids and consequent negative Ce anomaly of the carbonate precipitates (Kim et al., 2012;Pourret et al., 2008). The REE distributions of the fibrous coarsely crystalline facies from Ngol spring reveal an overall LREE enrichment (with LREE/HREE values as high as 3.69 in the non-normalised patterns, Table 3) and negative Ce anomalies ( Figure 12B,C). If the negative Ce anomaly can be interpreted as a signature typical of the inorganic precipitates from organic-rich, alkaline waters (cf. Pourret et al., 2008) the LREE enrichment suggests a contribution from deepseated crustal fluids (see discussion in Franchi et al., 2015). The deep, crustal origin of the parent fluids would be confirmed by the strong positive Eu anomaly that is often interpreted as a signature of the breakdown of plagioclase during hydrothermal activity (see below).
The REE distribution of the fibrous coarsely crystalline facies from Bongongo springs, on the other hand, is characterised by an overall slight depletion of LREE and a positive Ce anomaly suggesting a different scenario ( Figure 12B,C). If the overall LREE depletion is a typical signature of the inorganic precipitates from organic-rich, alkaline waters, the positive Ce anomaly might indicate either the solution of organic colloids during leaching of the samples or temporary anoxic conditions reached at time of deposition of these carbonates (Hu et al., 2014;Kim et al., 2012;Pourret et al., 2008 and references therein). The leaching of organic colloids co-precipitated with carbonate phases usually leads to a strong ∑REE enrichment (cf. Franchi et al., 2015Franchi et al., , 2016Pourret et al., 2008). The very low ∑REE in the fibrous coarsely crystalline facies samples from Bongongo rules out the organic particulate as potential source of the Ce enrichment. The trend in the studied samples is actually the opposite, with a negative correlation between ∑REE and Ce anomaly ( Figure 11E). This change in Ce partitioning from Ngol to Bongongo is coupled with an increase in Na and Mg concentrations ( Figure 11A; Table 3) and with an increase of the length-width ratio of the crystals probably due to higher CO 2 degassing (cf. Jones & Peng, 2012). These characteristics suggest that the Ce anomaly is bound to the nature of the spring fluids instead of the presence of organic colloids and might result from the mineralogy of the bedrock that is mainly granitic at Ngol (high overall REE contents) and basaltic (lower overall REE contents) at Bongongo. More insights on the behaviour of REE in this spring environment come from the comparison of the REE patterns in rapidly formed carbonates, that is fibrous coarsely crystalline microfacies, with slowly precipitated fine-grained carbonates, that is laminated filamentous-rich micrite from Ngol ( Figure 12D). The coarse calcite from Ngol has a similar REE distribution to the samples from Bongongo: very low ∑REE content and a slightly positive Ce anomaly (Table 3). From this comparison, it appears that the fast growth of carbonates caused by rapid degassing of CO 2 resulted in lower total REE incorporation into the carbonates, a lower La anomaly and higher Ce anomaly, while Eu and Gd seem to be unaffected. Hence, the Ce anomaly increases with rapid degassing, that is Ce gets complexed early in the fast precipitates with respect to other REEs. If the assumption that the coarsely crystalline calcite microfacies formed by rapid degassing is correct, then it becomes clear that, in continental settings, it is not only the intrinsic high water alkalinity and pH >8 that can promote Ce(III) oxidation and scavenging but more generally the flow velocity and the calcite precipitation rate. This fast growth mechanism in spring carbonates might explain extremely low REE contents and positive Ce anomalies detected in spring carbonates elsewhere and often interpreted as a result of organic matter contents and/or water chemistry (e.g. Great Artesian Basin of Australia, see Franchi & Frisia, 2020).

| Provenance of the spring water
The REE and trace element distribution in continental carbonates can be useful proxies for the characterisation of fluid/ rock interactions between the spring water and the rock substratum (Kokh et al., 2017;Teboul et al., 2016;Uysal et al., 2007Uysal et al., , 2009) and for the characterisation of the sources of calcium and other elements that constitute tufa and travertines (Kokh et al., 2017;Teboul et al., 2016). The application of carbonates as geochemical proxies for the provenance of spring water is a crucial tool when springs are dried or extinct.
The use of REE can be complicated by possible modification due to the presence of terrigenous matter characterised by high REE concentrations (Elderfield et al., 1990;Franchi et al., 2015Franchi et al., , 2016. The EPMA elements distribution maps of the carbonates from Ngol and Bongongo show the presence of small amounts of terrigenous materials mostly made up of Al-silicates (Figure 8). Although the weak acid leaching procedure should bring into solution only the carbonate phases, it is important to rule out possible REE contributions from silicates. The concentration of Rb can be used as an indicator of terrigenous contamination (Allwood et al., 2010;Franchi et al., 2016;Webb & Kamber, 2000). The data from Ngol and Bongongo show a weak correlation (lower statistical significance) between Rb and ΣREE, the La anomaly and Ce anomaly ( Figure 11G through I) suggesting negligible terrigenous contamination.
The HREE enrichment and slight LREE depletion in the PAAS-normalised REE patterns (corresponding to a LREE enrichment in the CL-Chondrite-normalised pattern, Figure 12B,C) and the low Th, jointly indicate precipitation from Ca-HCO 3 water and might indicate prolonged interaction of groundwater with carbonate aquifers (Choi et al., 2009;Kokh et al., 2017). Although this would agree with the water characterisation proposed by Fantong et al. (2019) for the area, the presence in the north-east of the study area (Ngol, see Figure 1C) of syn-tectonic and post-tectonic granitoids might explain the REE distribution found in the studied samples. Moreover, the CVL travertines are characterised by LREE (in the non-normalised data set) and Eu enrichment (Eu anomaly values >4; Table 3), which together suggest a contribution from deep-seated crustal fluids in contact with volcanic rocks and the breakdown of plagioclase (Douville et al., 1999;Franchi et al., 2015;Klinkhammer et al., 1994).
The marked positive Eu anomaly ( Figure 12; Table 3) in particular is indicative of the contribution of deep-seated crustal fluids in contact with the CVL alkali basalts (Kamgang et al., 2010(Kamgang et al., , 2013Njonfang et al., 2013). This model would be in line with the δ 13 C composition of the travertines from Bongongo and Ngol that points toward thermogene travertines and a deep-seated CO 2 source (Bisse et al., 2018).
Previous works have presented the chemical composition of the water in the Bongongo area (Lobe B soda spring in Fantong et al., 2019). The spring water presents relatively high (up to 37 mg/L) Fcontents of volcanogenic origin as the springs are located near the volcanically active Mount Cameroon (Fantong et al., 2019). The elevated 87 Sr/ 86 Sr ratio of 0.7133 yielded by the Bongongo spring water points toward the contribution of deep-seated water in contact with radiogenic lithologies occurring along the CVL (Aka et al., 2001;Fantong et al., 2019).

| CONCLUSIONS
The petrography and trace elements distribution, including REE, of carbonates from two locations, Ngol and Bongongo, along the CVL have been investigated. Based on the original data set presented here, and from water chemistry analyses available in the literature, the following conclusions can be drawn: 1. Petrographic analyses revealed two main microfacies: (a) fibrous coarsely crystalline calcite and (b) laminated filamentous-rich micrite and microsparite. The fibrous coarsely crystalline consists of millimetre to centimetrelong feather-like crystals of calcite formed predominantly inorganically under rapid CO 2 degassing in fast-flowing water along streams and cascades. The role of the microbial communities in this case was simply that of a catalyst for fabric development, that is vertical growth of the staked laminae fabric, making it a biologically influenced precipitate. 2. The laminated micrite microfacies, on the other hand, preserves an intricate mesh of hollow filaments up to 100 µm long and impregnated filaments. This fabric was probably formed in ponds or streams with slow-flowing water where the presence of microbial filaments exerted a dominant control on the growth of carbonates. Hence, although a biologically influenced precipitation of calcite is more probable, a biologically mediated origin for the laminated micrite microfacies cannot be excluded. 3. The carbonates from Ngol and Bongongo are characterised by PAAS-normalised REE patterns showing an overall LREE depletion and strong, consistent positive Eu anomaly (>4). The strong positive Eu anomaly and the δ 13 C values (from literature) suggest the contribution of deep-seated crustal fluids in contact with volcanic rocks and the breakdown of plagioclase, probably from the CVL alkali basalts. When compared with the samples from Bongongo the carbonates from Ngol have a higher ∑REE content. The discrepancy in the geochemistry of the travertines from the two localities highlights a difference in provenance for the spring fluids that can be explained with the contribution of water in contact with CVL granitoid and Mesozoic sediments. 4. This study has demonstrated how fluid velocity and carbonate formation rate can influence the geochemical composition of the precipitates. The strong CO 2 degassing, promoted by fast-flowing water, resulted in the rapid growth of coarsely crystalline calcite, characterised by overall ∑REE depletion and Ce enrichment. The fastforming calcite fabrics are therefore characterised by a higher Ce anomaly when compared to the slow forming fine grained, laminated fabrics. It can therefore be concluded that Ce(III) oxidation and scavenging into the coarse carbonates is promoted by fast-flowing water and strong CO 2 degassing (turbulence). This is an important finding for the interpretation of REE distribution in continental carbonates and provides an alternative explanation for the presence of a positive Ce anomaly that favours precipitation rates over organic matter contents.

ACKNOWLEDGMENTS
The authors wish to thank Thierry Bineli Betsi (BIUST) for facilitating the creation of this research team. Thanks are due to Tebogo Kelepile (BIUST) for the assistance at the EPMA. They thank Alexander Ziegler (University of the Witwatersrand) for the access to the SEM facilities. They are also grateful to the three anonymous reviewers for their useful suggestions that have contributed to greatly improve an early version of this manuscript. Petrographic and EPMA analyses were funded by the BIUST Initiation Grant R00010/2016 to FF.

AUTHOR CONTRIBUTIONS
SBB led the field work, devised the work and wrote the first draft of the paper, BEE participated in the field work and performed some of the analytical work, JG performed the SEM analyses, EE participated in the field work and provided supervision to SBB, FF performed the petrographic and EPMA analyses and led team in drafting the final version of the paper. All the authors have contributed in writing the paper.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.