The studied deposits in the Los Santos area (Figs. 2, 4, 5 and 6) are dominated by limestones over clastic sediments (i.e., consisting of extraclasts). The record showed a small proportion of polymictic clastic deposits composed of conglomerates, sandstones and occasionally mudstones, overlain by the dominant carbonate deposits that consist of varied sedimentary facies (Figs. 5 and 6). Three stratigraphic sections were measured (Fig. 5). These sections have been divided into several units for the sake of clarity in description. These units are based on lithology and/or textural characteristics. Individual facies description and facies labels in the following summary are found in Table 1.

A variety of sedimentary facies has been found (carbonate, detrital and peaty; Table 1 and Figs. 8, 9 and 10). The main mineralogy of the carbonate facies was calcite, as determined through microscopy and calcimetry analyses (see Materials and methods). In a few cases, these facies contain minor amounts of clay- to sand-size siliceous grains. A wide array of carbonate facies has been distinguished in the studied area based on texture and biotic composition (Table 1 and Fig. 8). The most abundant facies are the phytoclastic rudstones (Lph, Lphf), carbonate sands with gastropods and ostracods (Sb) and up-growing stem boundstones (Lst1). Less common are moss boundstones (Lbr), down-growing stem boundstones (Lst 2) and bioclastic limestones (Lb). In general, plant stems are coated with several concentric laminae (Fig. 11C, 11D), which in most cases contain bacterial evidence, i.e., cyanobacterial fan- and bush-shaped calcite bodies, as well as isolated filament moulds (Fig. 11E, 11F). Most of these facies were formed in multiple sub-environments, except for the occasionally occurring curtains of hanging stems, which formed in waterfalls. For example, up-growing stems developed in fluvial banks, shallow pools and floodplains; phytoclasts were ubiquitous, thus phytoclastic rudstones are found in almost every sub-environment; bryophyte layers formed in waterfalls, barrages and jumps.

Detrital (extraclastic) facies are characterized based on texture and sedimentary structures (Table 1 and Fig. 9). Conglomerates and gravels consist of angular to rounded, poorly sorted, carbonate and minor siliceous extraclasts, mainly from Mesozoic rocks, and at times including Quaternary tufa clasts, which together are embedded in a polymictic sandy-gravel matrix. They mostly form structureless deposits (Gm), but locally can show cross stratification (Gt). Sandstones and siltstones are also of polymictic nature, including variable amount of tufa intraclasts. All these clastic deposits occur at the base of the outcropping record and represent first incision and then deposition in low sinuosity channels during high energy events (Gm, Gt, St). Mudstones, consisting mostly of siliciceous grains (Fm, Fh), and marls (Mm) may be associated with thin peaty layers (facies C), giving evidence of periods of stagnant water on the floodplain and in ponds (Figs. 10; 11J, 11K).

The proposed sedimentation model was constructed based on the stratigraphic correlation and the various facies associations that resulted from progradation, aggradation and lateral migration of environments through time. The studied deposits represent the distal termination of a high-slope fluvial tufa system, in which the decreasing slope downstream favoured the development of palustrine areas, shallow ponded areas and small cascades. Moreover, the increase in width of the valley downstream favoured extensive development and preservation of the palustrine facies (i.e., up-growing calcite-coated stem boundstone). Therefore, the model corresponds to a low-slope, wide stretch dominated by palustrine conditions at the end of a stepped, cascade-barrage fluvial system (Fig. 13).

This model includes detrital sediments (extraclasts) that were deposited in shallow channels during incision periods due to high water discharge events, leading to gravel and sand deposits. Typically, channel incision during high discharge occurred at the initial stages of the Middle-Late Pleistocene record and was followed by gravel and sand fill and then by floodplain deposits, at places with accumulation of fine-grained carbonaceous matter (FA 1). Afterwards, carbonate sedimentation turned to be dominant throughout. Most flat areas became then site for palustrine vegetation with sluggish water flowing across extensive areas. The flooded action from some high-energy episodes eroded the up-growing stem phytoherms and associated deposits (e.g., carbonate sands) over the palustrine zones, producing phytoclasts that were deposited along the channels (e.g., at the beginning of FA 2 and 3) and on the palustrine areas. Some phytoclastic accumulations favoured the formation of small cascades and barrages-cascades along the channel, with deposition of fine sediment (i.e., carbonate sand and granule size) in the upstream dammed areas and moss layers downstream (FA 4). Progradation of these cascades and barrages-cascades structures gave rise to small hanging calcite-coated stem boundstones (e.g. as in FA 5). Hydrophilous vegetation and pooled areas with sandy sediment became dominant downstream in flat zones (FA 2 and 3), where aggradation of carbonate sand and lime mud in slow flowing and ponded areas occurred during low energy discharges (FA 2, 3 and 5). In this context, the hygrophilous plants were partially to fully submerged, then becoming coated with calcite and transformed into boundstone of up-growing stems. Also present is the abundant phytoclastic rudstone, which originated from the breakage of calcite-coated hydrophytes and broken stems accumulated in the palustrine environments. Water stagnation of some zones in these areas favoured accumulation of marly, silty and sandy sediment and preservation of peaty matter (FA 3).

Considering the large-scale sedimentary context of the Ebrón Valley, the different pre-Quaternary carbonate and non-carbonate rocks are incised by the present river as it flows down the decreasing slope gradient. There is a remarkable contrast between the upstream and downstream sedimentological features in the Pleistocene outcrops. The large cascade-barrage structures (formed of moss and stromatolite bioherms) and associated thick dam deposits upstream (consisting of carbonate sands and phytoclastic rudstones), e.g., near Castielfabib (as described in Lozano et al., 2012, and Sancho et al., 2015), became much less prominent downstream (e.g., near Los Santos). There, cascade-barrage structures were smaller and overall “small-slope and slow-water facies” were dominant (e.g., carbonate sands and silts, phytoclastic rudstones and stem phytoherms). Interestingly, stromatolites, which commonly form in fast-flowing water areas along moderate to high slope zones (e.g., Violante et al., 1994; Arenas et al., 2014b), are absent in the Los Santos area of the Ebrón Valley. Moreover, most moss mounds form discrete bodies that represent small jumps (Fig. 5). Another remarkable attribute in the Los Santos study area is the lack or limited presence of features denoting large or deep erosional processes within the tufa sequence, which supports deposition in overall slow to moderate flow condition.

These depositional changes coincide with a general decrease in slope and an increase in width downstream through the valley during the Pleistocene. This change was conditioned by variations of bedrock lithology, from Mesozoic carbonate rocks upstream to Neogene alluvial rocks downstream (e.g., as shown in Fig. 2). The gentle sloping topography of the Los Santos area and the lack of vigorous erosional processes at the distal termination of the fluvial system enabled a dominant palustrine condition to exist on the sluggish streams, along stream banks and in pools within wide floodplains.

The Los Santos tufa sequence is much thinner (up to 19 m) than the Castielfabib sequence (up to 77 m), which is consistent with the distal deposition from the carbonate source (springs upstream), leading to reduction of calcite precipitation downstream, due to the decrease of the dissolved calcium carbonate content and/or the diminishing mechanical CO₂-loss, related to lesser water turbulence. This feature has been shown in some modern fluvial tufa systems (Arenas et al., 2015). Remarkable is the fact that the present day water composition of the Ebrón River shows, mostly in the downstream area, minor decreasing trends in Ca content and alkalinity values, but the values of the Saturation Index with respect to calcite usually remains higher than +0.6, suggesting the existence of tufa sedimentation in the distal part of the river, though reduced compared to upstream deposition rates (Arenas et al., 2015). Moreover, present day sedimentation studies indicate that the tufa deposition rates in each fluvial subenvironment of the Ebrón River were mainly controlled by the CO₂-outgassing intensity linked to local flow conditions and the biological substrate type (Arenas et al., 2015). In the Pleistocene study case, a decrease in CO₂-outgassing intensity would be expected, as the area corresponded to the downstream, low-slope area. As a matter of the fact, thickness of deposits on tablets monitored every six months along that river also showed decreasing depositional rates downstream, coinciding with the area of Los Santos. Additional water inputs through springs that introduce CO₂ in the carbonate system may cause decreases in calcite precipitation, as in the case of the present Ebrón River (Arenas et al., 2015). This type of water input was occasional in the downstream stretch. In the Pleistocene, although spring CO₂-inputs may have accounted for the small tufa accumulation at the downstream part of the system, this type of inputs is not expected to have been dominant, given the extensive palustrine tufa deposits. Thus, springs, if present, were small or occasional.

Many other ancient sedimentary fluvial systems with slope and facies changes along the valley path have been found in the Quaternary record (e.g., Violante et al., 1994; Pedley et al., 1996; Arenas et al., 2014a; Capezzuoli et al., 2014; Huerta et al., 2016; Toker, 2017). Several other Quaternary examples with palustrine environments come from central Italy (Buccino et al., 1978), Israel (Heimann & Sass, 1989), central Spain (Pedley et al., 2003), southern Turkey (Özkul et al., 2010) and southwestern Tunisia (Henchiri, 2014). However, situations with extensive low-slope palustrine tufa areas at the most downstream fluvial stretch, as in the case study of Los Santos, are not so commonly preserved.

Arenas et al. (2014a) described a moderate-slope fluvial tufa model for Holocene deposits in the Iberian Ranges in which some portions with dominant palustrine facies could be similar to those in the Los Santos studied example. Thick palustrine deposits have also been described in the lower reach of the Pleistocene Mesa River (Váquez-Urbez et al., 2012). Henchiri (2014) related the dominant palustrine conditions of Holocene tufas in southwestern Tunisia to bedrock configuration and the flat topography of the area where the springs emerged, prone to form a marshy context.

In the Los Santos studied system, the large dammed areas, opposed to minor and small cascades-barrages, formed as a result of gentle gradient in the wider valley, thus vertical growth of barrages was limited, while favouring lateral migration of the fluvial environment. Stagnant conditions in the pooled areas could be established due to the likely long periods without water renewal and lack of vigorous flow, which together favoured accumulation and preservation of organic matter such as peat (e.g. as described by Pedley et al., 2003; Arenas et al., 2014a; Henchiri, 2014).

The palustrine tufa of the Ebrón River is a typical example illustrating the distal evolution of a carbonate-rich riverine environment in a wide low-layering area that, in the downflow direction, progressively reduces or stops its capacity to precipitate calcite due to the decrease of the dissolved calcium carbonate and/or the diminishing mechanical CO₂-loss, related to lesser water turbulence in overall low-slope surfaces. In other settings, this calcite precipitation reduction has been related to the system entering other hydrological system with different hydro-geochemical characteristics (see Toker, 2017). In these settings, carbonate deposits are rapidly replaced by detrital tufa or mixed clastic successions (Ortiz et al., 2009; Martini & Capezzuoli, 2014). However, this is not the case of the Pleistocene example studied herein.

Together, the above discussion indicates the complex interplay of factors that control fluvial tufa deposition (geology, topography, hydrology, hydrochemistry, biology, climate; see Goudie et al., 1993; Pedley et al., 2003; Pentecost, 2005; Valero Garcés et al., 2008; Arenas-Abad et al., 2010). As cited above, the factors that favour the formation of tufa palustrine systems are: the reduced slope and large amplitude of the sedimentary substrate, the high saturation of water in calcium carbonate and the reduced and slow water level changes, along with the absence of strong erosional processes.

Subsidence is a principal factor that favours accumulation and preservation of different types of deposits. In the investigated environment, for instance, thick tufa, oncolite-bearing and bioclastic limestones formed in the middle-late Miocene of the Ebro Basin; thickness and preservation were clearly linked to subsidence induced by evaporite-solution of underlying bedrocks (Arenas et al., 2000). In this case, the corresponding beds showed typical geometry and thickness variations. Similarly, thick and varied facies deposits formed westward in the Ebro Basin, in relation to subsidence associated to a blind thrust rooted in the Iberian Range during the middle-late Miocene (Vázquez-Urbez et al., 2013). However, thick dominantly palustrine tufa successions preserved in basins affected by tectonic activity and subsidence have been reported only in a few cases (Ellera Basin; Pazzaglia et al., 2013).

Due to the concurrence of such varied conditions, extensive areas of palustrine tufa at the downstream areas of fluvial systems form only rarely and related to local situations. For instance, in arid conditions (Ben Younes tufa - Gafsa area, Tunisia) Henchiri (2014) described extensive palustrine tufa deposits related to repeated occurrence of springs able to lead supersaturated waters. Tufa deposits can also represent the distal portion of thermal systems, where typical travertine facies are transitional to cooler, plant-rich tufa facies and especially of suitable environments as lakes, alluvial plains or marshes (Della Porta, 2015; Della Porta et al., 2017; Brogi et al., 2017; Mancini et al., 2019; Mohammadi et al., 2020).

In brief, from the above discussion, the factors that favour the formation of tufa palustrine systems are: the reduced slope and large width of the sedimentary substrate, the water saturation in calcium carbonate and the reduced and slow water level changes, along with the absence of strong water discharge provoking erosion. Preservation is expected to be greatly affected by subsidence, as in any other subaerial contexts. Whether subsidence could control the characteristics of the studied palustrine system in the Ebrón valley is unknown.