Quaternary fluvial carbonate deposits of the Almonda River Valley, Central Portugal

This paper discusses the formation and preservation of a fluvial tufa system influenced by Atlantic climate based on stratigraphical, chronological (amino-acid racemization, AAR), sedimentological and stable-isotope analyses. On the southwestern Iberian Peninsula, the tufas and associated deposits of the Almonda River valley occur as isolated terraced bodies and reach 25 m thick. AAR dated most deposits to within the warm Marine Isotope Stage 5 (MIS-5). Two Holocene ages were reset within MIS-5 based on diverse criteria. Widely varied carbonate and minor allochthonous coarse detrital facies occur arranged in four simple vertical associations. The deposit geometry and facies association distribution correspond to a low- to moderate-sloped fluvial valley consisting of several short knickpoints and extensive flat areas between them. The latter are occupied by slow-flowing water facies (carbonate sand, lime mud, phytoclast and oncoid rudstones, and up-growing stem boundstones). Facies that formed in moderate- to high-slope substrates were stromatolite, moss and down-growing stem boundstones. The homogeneous Miocene bedrock lithology and gentle structural deformation propitiated this depositional architecture. Calcite δ13C and δ18O values suggest that the aquifer water provided the outflowing Almonda water with (1) 18O-enriched water, compared with present precipitation and groundwater δ18O values, and (2) 13C-depleted CO2 from bituminous rocks and vegetation cover in the catchment. The proximity to the Atlantic coast favoured the Mesozoic-rock aquifer recharge with 18O-enriched water precipitation, assuring water availability during the formation of the studied tufas. No evidence of frequent intense erosion phases might indicate stable precipitation regimes, which would have allowed the preservation of loose fine-grained and palustrine deposits.


Introduction
Tufas are calcium carbonate rocks, mainly calcite, that form in terrestrial environments, such as streams, lakes and springs (Capezzuoli et al., 2014;Pedley, 1990). Freshwater carbonate-depositing rivers and lakes are the most common contexts for tufa formation, but saline and alkaline lakes may also be loci for tufa formation (Della Porta, 2015;Pentecost, 2005;Renaut et al., 2013). These limestones result from calcite precipitation on diverse biotic substrates, primarily through CO 2 loss from water caused by changes in temperature, biological processes and/or mechanical agitation (Pentecost, 2005). Tufa-associated carbonate sediments are commonly included within the term "tufa", as is the case for oncolites and stromatolites.
In mid-latitude regions, tufas developed extensively through the interglacial Quaternary periods (i.e., time periods correlative to odd marine isotope stages, MISs; e.g., Gibbard et al., 2005), which were characterized by warm conditions, with prominent deposits at approximately 100 ka (corresponding to MIS 5), as shown by the works of Martín Algarra et al. (2003), Torres et al. (2005), Ortiz et al. (2009) and Sancho et al. (2015). However, tufa formation has been found to occur in more varied climate conditions, ranging from temperate-wet to semiarid and arid (Cremaschi et al., 2010;Kele et al., 2021;Moeyersons et al., 2006;Stone et al., 2022;. In fact, some tufa deposits in the Iberian Range formed during MISs 8 and 6 (Torres et al., 2005;Valero-Garcés et al., 2008;Sancho et al., 2015)two "glacial" periods-. Together, these facts suggest that factors other than climate can also influence tufa formation (Viles & Pentecost, 2007), such as the bedrock lithology, the characteristics of the related aquifer, and basin slope variations along water flow directions (Arenas-Abad et al., 2010). Moreover, intrinsic factors, such as flora development and physical and chemical characteristics of water, may play a crucial role in some characteristics of tufa deposits, such as the type of facies and the geometry and thickness of the deposits (Arenas et al., 2014a(Arenas et al., , 2014bArenas-Abad et al., 2010;Violante et al., 1994).
Some studies have focused on the palaeoclimatic significance of tufas because their formation appears to be favoured in warm (Durán, 1989;Henning et al., 1983;Martín Algarra et al., 2003;Ortiz et al., 2009) and/or humid conditions (Capezzuoli et al., 2010). However, it is important to note that most of the abovementioned climate-related contributions are based on knowledge of the age of the deposits, but for the most part, the tufa attributes per se do not allow us to decipher the precise climate conditions during their formation. However, in a few cases, climate conditions have been described based on particular features of flora and/or fauna in tufas (Martínez-Tudela et al., 1986;Ricci et al., 2015;Roiron, 1997;Rubio-Millán, 2002;Bertini et al., 2014;Aranbarri et al., 2021).
There are few known examples of the influence of climate on tufa formation during Quaternary interglacials in the European region influenced by Atlantic climate. Except for a few notes describing tufas in Portugal (e.g., Choffat, 1895;Flyche, 1907;Mendes, 1974;Soares et al., 1997) or as part of other geomorphological studies (e.g., Santos, 1996), the first works considering sedimentary, geochemical and/or paleontological aspects with hydrological and climatic implications in central Portugal are by Queiroz and Mateus (2011) and Ribeiro et al., (2013Ribeiro et al., ( , 2014. To the south of the study area, Guerreiro ( , 2015 and Guerreiro et al. (2010) described different tufa facies in Quaternary deposits in the central Algarve. Facies associations were interpreted in relation to geomorphological features and water flow conditions, fitting the stepped fluvial models of Arenas-Abad et al. (2010). Pedley et al. (1996) compared Holocene barrages from Spain and Britain and suggested that the "different characteristics of their preserved fabrics should allow recognition of two contrasting climatic models for fluvial barrage tufas in ancient deposits". Azennoud et al. (2021) demonstrated the response of tufas to climate on the multi-centennial time-scale over the last 5000 years in northwestern Africa.
This work presents the results of a study of Quaternary tufas and tufa-associated deposits in the Almonda River valley (central part of Portugal; Fig. 1A, B) based on stratigraphic, chronological and sedimentological analyses, along with stable isotope analysis of the carbonate facies. The main objectives were (1) to interpret the depositional stages based on stratigraphic and numerical dating criteria, (2) to characterize the depositional system through facies analysis and provide a sedimentary facies model, and (3) to discuss the factors that influenced the formation of the carbonate and associated deposits in that region. This is the first study that assesses the depositional, hydrological and climatic significance of fluvial tufas in this region, providing new information that can help sedimentological and climatic interpretations elsewhere. Moreover, this study is intended to highlight the potential of tufa as an educational and touristic resource in the Iberian Peninsula.

Geographical and geological location
The study area is located in the central part of Portugal, 95 km to the northeast of Lisbon, in an area known as Médio Tejo or Ribatejo (Fig. 1A). The studied Quaternary tufas and associated deposits crop out along the WNW-ESE-oriented stretch of the Almonda River valley (Fig. 2). The Almonda River is a direct tributary of the Tagus River. It flows from west to east for 13.8 km and then turns and flows north to south for 20.4 km. The altitude range from 75 m above sea  Santos, 2010) level (asl) at the uppermost spring to 28 m asl at its joining with a northflowing stream (Alvorão valley), close to the locality of Torres Novas. The Almonda River enters the Tagus River 20.4 km downstream of Torres Novas at 15 m asl (Fig. 1B).
The present climate of the region is temperate, with rainy winters and dry, mild summers upstream of the source of  Manuppella et al., 2000) and n. 27 -C (Torres Novas;Manuppella et al., 2006), of the Instituto Geológico e Mineiro of Portugal the Almonda River (Estremenho Limestone Massif), and with rainy winters and dry, hot summers downstream of the Almonda source (https:// www. ipma. pt/ pt/ educa tiva/ tempo. clima/ index. jsp? page= clima. pt. xml).
Data from Santarém (30 km south of Torres Novas in the Tagus Basin; Fig. 1A) provide the maximum average temperature of the hottest month of summer around 30 °C and the minimum average temperature of the coldest month of winter around 6 °C. It is normal to have maximum temperature peaks in the summer above 40 °C (Instituto Português do Mar e da Atmosfera, I. P., IPMA, I. P.; Fino, 2018).
The average precipitation in the region is higher in autumn and winter than in spring, with minima in summer. The average monthly data compiled from the Ribatejo region by Fino (2018) indicated that precipitation ranged between 295 mm (autumn), 220 mm (winter), 125 mm (spring) and 45 mm (summer) since 2006-2014. Most of the river path is meandering (Manuppella et al., 2000) (Fig. 2). The alluvial plain can reach up to 300 m wide. In four sections of the river, the meanders tend to form abandoned stretches (oxbow lakes).
The mean river slope along the WNW-ESE stretch is approximately 0.34%, although it is highly variable along the valley. It increases significantly from kilometre 8 to the Torres Novas locality, reaching up to 5.5% between Lapas and Torres Novas. From Torres Novas, the river flows downhill with an average slope of 0.06%, as far as the Tagus River (Fig. 3). There are three knickpoints along the river: immediately after the source, between Lapas and Torres Novas and in Torres Novas. However, these points may have been changed by the recent construction of dams and weirs.
Jurassic limestone rocks crop out to the northwest in the Estremenho Limestone Massif (Figs. 1B, 2), a geomorphological unit with an extent of approximately 800 km 2 , and this massif is high above the Cenozoic Tagus Basin. These rocks constitute an important and complex aquifer system with typical karst behaviour. The Miocene rocks of the Tagus Basin consist of limestones, marly limestones and marls, with sandstone and mudstone intercalations corresponding to lateral facies variations (Manuppella et al., 2000). These rocks are widely represented to the east and south of the Jurassic outcrop. The contact between these two geological units is a fault (Arrifes thrust), through which the two main perennial karst springs from the massif, namely, the outflows of the Alviela and Almonda Rivers, are located, with average values of water flow from the karst system of 120 hm 3 /year and 80-100/hm 3 /year, respectively . The chemical composition of the river water is of the calcium bicarbonate type Manuppella et al., 2006).
The Quaternary tufa deposits that crop out along the WNW-ESE stretch of the Almonda River valley and some of its tributaries, from the upstream springs to the locality of Torres Novas, are studied in this paper. These deposits unconformably overlie the Miocene bedrock and appear incised by the present Almonda River (Figs. 2, 3). The contact between the Miocene bedrock and the tufa and associated carbonate and detrital deposits is not always observable. These tufa deposits crop out as terraced isolated build-ups along the present-day Almonda River valley and a few isolated deposits in one lateral tributary valley (Figs. 2,3). Some close tufa outcrops correspond to Fig. 3 Longitudinal profile of the Almonda River and tufa altitude across a WNW-ESE section (from the upstream spring to the south of Torres Novas town) adapted from Thomas (1985). The Vale Sobral tufa creek on the north bank (VSO) is also projected. The location is shown in Fig. 2 the same build-up that has been dissected by the present river (e.g., the Ribeira Ruiva and Ribeira Branca outcrops) (Fig. 2). Tufa deposits occur approximately 2.5 km from the main water source, which is located at the contact with the Jurassic rocks. The altitude of the base of the outcropping deposits ranges from 59 m (terrace close to Casal da Pinheira) to ca. 28 m (terrace of Moinho da Cova, in Torres Novas) (Fig. 2).
The Lapas locality is built on an extensive tufa buildup, whose interior observation is possible through a series of artificial tunnels excavated in the tufa in the past. The tunnels occur in the middle part of the Lapas build-up, with vertical conduits (chimneys) that reach the surface at an elevation of 60 m asl. The lowest tunnels were opened at an elevation of 50 m asl. The labyrinthic transect spans approximately 400 m long. These tunnels and other close tunnels in the area have been considered of public interest since 1943, and have formed part of the Roteiro das Minas e Pontos de Interesse Mineiro e Geológico de Portugal since 2019.

Materials and methods
Tufa and associated deposits in the study area were mapped on a 1:10,000 topographic basis by superposing information of the 1:50,000 geological sheets n. 27-A (Vila Nova de Ourém; Manuppella et al., 2000) and n. 27-C (Torres Novas;Manuppella et al., 2006) (Fig. 2). Three stratigraphic sections (ACHOP-Alto do Choupal-, MOICO-Moinho da Cova-, and RIB-Ribeira Branca-) were measured and located on a 1:10,000 topographic basis. Several other outcrops were also studied. Correlation between the several outcropping deposits was based on mapping, lateral continuity of outcrops and the ages provided by amino acid racemization (explanation given below). Sedimentological drawings of diverse outcrops were also made based on photographic mosaics. From the measured sections and several other outcrops, 42 samples were selected based on their stratigraphic position and lithofacies types for diverse types of analyses, including their textural, mineralogical and stable isotope characteristics, and dating. Twenty-five thin sections were prepared for the textural characterization through petrographic microscope observations, and 8 samples (whose approximate dimensions are 5 × 5x3 mm) were prepared for scanning electron microscopy (SEM). The thin sections were made at the Servicio General de Preparación de rocas y materiales duros, impresión y escaneado en 3D of the Servicio General de Apoyo a la Investigación (SAI) of the University of Zaragoza (Spain). The SEM facility was a Carl Zeiss MERLIN™ FESEM; Zeiss AG, operating at 3 to 15 kV and 158 pA, which belonged to the Servicio de Microscopía electrónica de materiales of the SAI.
The classification codes of Miall (1978) have been used for the clastic facies (i.e., rocks formed of extraclasts), and the classification codes of Arenas-Abad et al. (2010) have been used for the carbonate facies.

Amino acid racemization (AAR)
A total of 9 samples from 9 beds corresponding to carbonate silts and sands located in six different tufa deposits (Fig. 4) were used for amino acid racemization (AAR) dating in the Biomolecular Stratigraphy Laboratory at the Polytechnic University of Madrid.
A total of 17 ostracod valves from 7 beds were analysed (Table 1). The valves were carefully sonicated and cleaned with water to remove sediment. They were then left for 3 h in hydrogen peroxide to eliminate organic matter. Only clean and translucent shells were selected from each sample. In addition, a small sample near the aperture of gastropod shells belonging to the Theba genus (Helicidae family) was recovered in 4 specimens from 2 samples (Table 2). In all cases, peripheral parts (approximately 20-30%) were removed after chemical etching with 2 N HCl. Afterwards, 10-20 mg of sample was selected for amino acid racemization analysis.
Amino acid concentrations were quantified using highperformance liquid chromatography following the sample preparation protocol described in Kaufman and Manley (1998) and Kaufman (2000). This procedure involves hydrolysis, which was performed under a N 2 atmosphere in HCl for 20 h at 100 °C. The hydrolysates were evaporated to dryness in vacuo and then rehydrated in 7 mL (ostracodes) or in 10 μL/mg (gastropods) of 0.01 M HCl with 1.5 mM sodium azide and 0.03 mM L-homo-arginine (internal standard).
Samples were injected into an Agilent HPLC-1100 liquid chromatograph equipped with a fluorescence detector. Excitation and emission wavelengths were programmed at 230 nm and 445 nm, respectively. A Hypersil BDS C18 reverse-phase column (5 μm; 250 × 4 mm i.d.) was used for the analysis.
The age of a single bed is the average of the numerical dates obtained for each amino acid D/L value measured in gastropods or ostracods from that level, and the age uncertainty is the standard deviation of the numerical ages calculated at each level. The D/L nomenclature refers to the position of the NH 2 group to the right (D isomer) or to the left (L isomer).

Stable isotope analysis
A total of 14 samples representing diverse facies from two sections were obtained through a microdrill and a punch and then ground and sieved to 62 μm, for C and O stable isotope analyses. δ 13 C and δ 18 O were determined at the Servicios Científico-Técnicos (CCIT-UB Serveis) of the University of Barcelona (Spain). The isotopic ratios were measured in a mass spectrometer (MAT-252, Thermo Finnigan; Thermo Fisher Scientific, Waltham, 1 3 MA, USA) via CO 2 obtained in a Carbonate Kiel Device III (Thermo Finnigan) by the reaction of samples with 100% H 3 PO 4 at 70 °C for 3 min (McCrea, 1950). The international standards NBS-19 was used (δ 13 C = 1.95‰; δ 18 O = − 2.20‰ V-PDB). The overall reproducibility was better than 0.02‰ for δ 13 C and 0.04‰ for δ 18 O. The results are reported in δ‰ notation relative to Vienna Pee Dee Belemnite (V-PDB) ( Table 3).

Amino acid racemization results
To establish the numerical ages of the ostracod samples used for AAR, aspartic acid (Asp) and glutamic acid (Glu) were selected because they account for a significant percentage (> 50%) of the amino acid content of most ostracod valves (Kaufman, 2000;Ortiz et al., 2013). Other amino acids were not considered due to their low racemization rates. The mean Asp and Glu D/L values in ostracod valves from the samples are shown in Table 1.
Of the ostracod samples, 8 results were excluded because the Asp and Glu D/L values were off of the covariance trend (cf. Kaufman, 2003Kaufman, , 2006 and/or because of abnormally high D/L values, characterized by Asp D/L and Glu D/L values plotting outside the 2σ range of the group (cf. Hearty et al., 2004;Kosnik & Kaufman, 2008). Specifically, we excluded 1 MPAU sample; 1 LAP 02 sample; 2 RIBR 01 samples; 1 TNMC 01 sample; 2 LPCM 01 samples, and 1 TNMC 03 sample.
For dating ostracod samples, we applied the age calculation algorithms described by Ortiz et al. (2004) for samples with Asp and Glu D/L values greater than 0.27 and 0.08, respectively. For the age calculation of ostracods of samples MPU and TNMC 03, we used the age calculation algorithm defined by Ortiz et al. (2015) for samples with Asp D/L values lower than 0.30 (Holocene samples): age (yr) = 17.74 D/L Asp -0.90. In this latter case, we considered only Asx because it is the amino acid that racemises the fastest, and D/L Glu values are not able to discriminate Holocene ages .  To establish the amino chronology of the Helicidae shells from the carbonate deposits, we used the Asp, Glu, leucine (Leu), and isoleucine (Ile) contents, according to Goodfriend (1991). The analysis of more than one amino acid provides largely redundant information on the sample age. The mean Asp, Glu, Leu and Ile D/L values at each tufa level are shown in Table 2 together with their ages.
The numerical age of the Theba shells at each level was calculated by introducing the Asp, Glu, Leu and Ile D/L values into the age calculation algorithms established by Torres et al. (1997) for gastropods in the central and southern parts of the Iberian Peninsula.

Stratigraphy
The studied Quaternary record overlies Miocene rocks, although in some cases the base of the Quaternary deposits does not crop out. Figures 5, 6, 7, 8, 9 present three representative sections (Alto do Choupal, Ribeira Branca and Moinho da Cova, abbreviated as ACHOP, RIB and MOICO, respectively, Figs. 5, 6 and 9), along with field images of some conspicuous characters. The studied deposits are dominated by limestones and carbonate sands and silts, which, in some cases, are found to lie over less abundant clastic deposits formed of conglomerates and gravels (e.g., Alto do Choupal, ACHOP, Fig. 5A, F). The thicknesses of some build-ups and sections vary between a few metres (e.g., in the Vale Sobral, VSO section, and Ribeira Branca, RIB section, Figs. 2, 6, 8) and up to 23 m (e.g., in Moinho da Cova, MOICO section, Fig. 9A), although some build-ups can be thicker than the measured sections, e.g., at Ribeira Branca, up to 23 m thick (Fig. 3). All outcropping deposits are physically disconnected. Thus, the relative stratigraphic position of these deposits can be established only by combining geological mapping and absolute dating. The ages obtained through AAR in several silty and sandy carbonate deposits herein allow us to separate two chronological groups of deposits (Tables 1, 2): 1) Pleistocene deposits. Most tufa build-ups in the study area coincide with the time of MIS 5 (i.e., Middle-Late Pleistocene). Precise stratigraphic correlation between the several Pleistocene build-ups and between the measured sections has not been possible (Figs. 5,6,9), as the obtained AAR ages overlap.
2) Holocene deposits. These deposits include the uppermost part of the MOICO build-up over an erosional surface, with an age of 8.97 ka, and the MPAU build-up (sample taken close to the base of the section), with an age of 1.35 ka. However, the AAR ages of these two build-ups are not fully consistent with their geomorphological and topographical positions relative to the surrounding dated deposits.
Below is a summary of the main stratigraphic characteristics of the most representative build-ups, including chronological considerations of some of them. Individual facies description and facies labels in the following summary are found in Table 4.

Middle-Late Pleistocene deposits
These deposits crop out along the main valley and represent the majority of the Quaternary tufa record. Although not all deposits could be dated due to the lack of suitable samples or due to the above-described technical problems relative to the AAR method, their topographic and stratigraphic positions relative to other dated tufa build-ups and alluvial terraces allowed us to tentatively assign them an approximate age. This is the case of the Alto do Choupal build-up (ACHOP), which is ca. 15 m thick, along the middle stretch of the Almonda valley. The proximity and position of this build-up relative to the dated Ribeira Branca (RIB) and Ribeira Ruiva (RIVR) build-ups ( Fig. 2) allowed us to place it within MIS 5. Figure 5 shows a stratigraphic section and field images of the ACHOP outcrop. Polygenic conglomerates crop out at the base, on Miocene marls (Fig. 5A, F, G).
The Ribeira Branca and Ribeira Ruiva build-ups, which are set in front of each other and separated by the present river, are 23 and 22 m thick, respectively (Figs. 2 and 4). The former yielded two close ages, 116.2 ± 28.37 and 124.38 ± 20.4 ka, and the latter yielded 112.4 ± 12.1 ka (Tables 1 and 2). Figures 6 and 7 show a stratigraphic section and illustrate the outcrops. Both lie on Miocene beds: coarse carbonate sands containing dispersed millimetre-to centimetre-long clasts in the northern RIB outcrop (Fig. 7A, B) and conglomerates at the base of the RIBR build-up occur on Miocene nodulised marly-limestones ( Fig. 7A, B, C).
The deposits over which the Lapas locality is located (Lapas terrace) are another example of Pleistocene tufas. This build-up is up to 16 m thick. It stands out because of the extensive and intricate development of elongate caves (i.e., tunnels) carved mostly in fine-grained carbonate-dominated deposits, i.e., composed of fine phytoclasts (Lphf), small oncoids and sands (Lo, Sb), and stem boundstones (Lst 1) (Fig. 7E, F) (Fernandes, 2018). The Lapas cemetery terrace, to the northwest of the Lapas terrace, is a 16 m-thick buildup composed of fine phytoclasts (Lphf), small oncoids and carbonate sands (Lo, Sb). Dating of a carbonate sand sample provided an age of 121.1 ± 14.7 ka ( Fig. 2 and Table 1).

Pleistocene or Holocene deposits?
A small wedge-shaped deposit, which is 1.5-2 m thick, along a tributary of the Almonda valley (Vale Sobral; VSO, Figs. 2, 3) is mainly formed of stromatolite and speleothem laminated facies ( Fig. 8A-C). It is tentatively set within MIS 5 due to its topographic position, but a different age could also be possible for the VSO.
At the Torres Novas locality, the MOICO build-up occurs on both banks of the present Almonda River (Figs. 2, 3). The MOICO section is 20.1 m thick and includes several nonexposed stretches (Fig. 9A). Close to the section's base, the lowermost tufa deposits in the sector almost reach the course of the present river, adding approximately 2 m to the total thickness in the MOICO section. Figures 9 and 10 illustrate the outcrop.
Dating the MOICO build-up is not straightforward. The only available age is from a sample of the uppermost deposit over an erosional surface (Fig. 4F); sample TNMC 03 yielded 8.97 ka, which is early Holocene (Table 1). Another sample taken for AAR dating (TNMC 01, Table 1) that is ca. 2 m above the course of the present river (approximately equivalent to metres 4-6 of the MOICO section; Fig. 9A), has not given a reliable age. Thus, the MOICO deposit below the erosional surface can correspond to any age older than 8.97 ka.
Interestingly, the alluvial terrace in the lowest reach of the Almonda River at the confluence with the Tagus River is the Tagus youngest terrace identified by Martins et al. (2008) and Martins & Cunha (2009). Dating this Tagus alluvial terrace (named T6 by those authors) by Infrared Stimulated Luminiscence (IRSL) at Azinhaga allowed the deposition of this terrace to be estimated to between 32 and 62 ka, i.e., MIS 3 (Cunha et al., 2016). Terraces T5 + T6 are identified to be equivalent to previously mapped terrace Q4 (8-15 m thick) in geological sheet 27-C (Manuppella et al., 2006) (Fig. 2). Q4 deposits appear at a lower topographic position and are inset into the MOICO build-up. Poorly exposed siliciclastic mudstones and gravel deposits that occur at an approximate position between metres 8 and 12 of the MOICO section are attributed to terrace Q4 herein. If the above considerations are correct, the MOICO buildup below the erosional surface should be older than MIS 3. The climate during MIS 4 was not favourable for tufa formation, as shown by the rare tufa record of most pairnumbered MISs (Martín Algarra et al, 2003;Sancho et al., 2015). Thus, an age within MIS 5 is tentatively proposed for the MOICO build-up, coinciding with the majority of deposits in the study area.
The uppermost deposits of the MOICO build-up, separated from the phytohermal bodies below through an erosional surface, are up to 1.5 m thick. These deposits consist mostly of phytoclast and intraclast rudstones with carbonate sand patches (where sample TNMC-3 was taken) and are poorly exposed and laterally discontinuous (Figs. 4F, 10E, F). The AAR dating placed them within the early Holocene (8.97 ka, Table 1). However, this age is not consistent with the high elevation of the deposits over the likely MIS-5 MOICO deposits. Moreover, the fact that the sample was taken from mixed-up sediments close to a human-made wall suggests that caution should be taken when considering the Holocene date. It is possible that the MOICO's uppermost deposits are closer in time to the underlying MOICO deposits.
The MPAU build-up crops out along the middle stretch on the right bank of the present Almonda River (Figs. 2, 3). The base is not observable; the outcropping deposits are 23 m thick. The top forms flat surfaces that at present are used for agricultural activities. The outcropping deposits are dominated by carbonate sands and silts, followed by stem boundstones and phytoclast rudstones (Figs. 4C, 11). The geomorphological and topographic positions of the MPAU build-up suggest that this deposit should have formed at the same time as others of the same elevation, as is the case for the LPCM, dated at MIS 5 (Table 1) and LAP deposits (Figs. 2, 3). Therefore, the MPAU build-up is tentatively set within the Pleistocene group, perhaps within MIS 5.

Sedimentary facies
The sedimentary facies (i.e., lithofacies, as defined on the basis of the lithology and physical and biological features) present in the study area are diverse and have been described in other fluvial tufa-bearing systems (e.g., Arenas et al., 2000;Arenas-Abad et al., 2010;Azennoud et al., 2021;Capezzuoli et al., 2009;Ordóñez et al., 2016;Pedley, 1990Pedley, , 2009Rodríguez-Berriguete et al., 2018Sallam, 2022;Vázquez-Urbez Fig. 5 Stratigraphic section of Alto do Choupal (ACHOP) and related field images. The legend of symbols in A are given in Fig. 14B. B Boundstone consisting of inclined moss layers. C A small cavitiy with speleothemic overgrowth over down-growing stems (Lsp). On the right, vertical layer of moss boundstone Lbr). D Stromatolite dome. E Domed deposits consisting of highly inclined layers of moss boundstone, stromatolites and speleothems. Flow direction towards the left (southeast). F Phytoclast and minor oncoid rudstone (calcite coatings), including minor extraclasts. Note the oncoid to the left of hammer (arrow). G Detail of F, showing the fining-upwards evolution of the clast sizes ◂ et al., 2012; Violante et al., 1994, among others). Herein, there is no distinctive difference between facies characteristics in the different build-ups; thus, no separation has been made when describing the facies. The dominant sedimentary facies are those consisting of carbonate sediments, followed by minor conglomerates and gravels. An array of carbonate facies has been distinguished based on geometry, textural features, sedimentary structures and biotic components ( Table 4). The main mineralogy of all carbonate facies is calcite, as determined through microscopy analyses (both in optical and electron microscopy). The most abundant facies are the phytoclastic rudstones (Lph and Lphf), carbonate sands and silts with  Fig. 14B. B Field view of the section, with dominant carbonate sand (Sb) and fine phytoclasts (Lphf), followed by facies Lst 1. C Downstream-inclined layers consisting of phytoclast and fine phystoclast rudstone, followed by fences formed of up-growing stem boundstone (Lst 1). D Lime mud and silt containing intraclasts (In) and gastropods (Ga). E Detail of B showing interbedded brown-coloured lime silts. F Detail of up-growing stem boundstone and some low-lying horizontal phytoclasts. Note the laminated calcite coatings (arrow). As per the inner diameter, the stems could correspond to cane-like vegetation gastropods and ostracods (Sb) and stromatolites (Ls), followed by the moss boundstones (Lbr) and the up-growing stem boundstones (Lst 1). Less common are the down-growing stem boundstones (Lst 2), the oncolite rudstones (Lo) and the bioclastic and intraclastic limestones (Lb and Li). Speleothem facies (Lsp) are rare. A majority of calcite coatings on plant stems from either Lst 1, Lst 2 or Lph consist of several concentric laminae (Fig. 12A), which in most cases contain bacterial evidence, i.e., filamentous cyanobacterial calcite bodies arranged in fan-and bush-shaped bodies and isolated filaments (Fig. 12B, C). Cyanobacterial evidence is also found in stromatolites, oncolites and certain laminated speleothem deposits ( Fig. 13A-F). Calcified moss caudilia and filidia can also contain bacterial evidence (Fig. 13G, H). Table 4 summarizes the most important macroscopic and microscopic characteristics of every facies, as well as their abundance, the facies that are commonly related to each other and their depositional environment interpretation. With the exception of the curtains of hanging stems (Lst 2) and speleothems (Lsp), most of these carbonate facies could have developed in several (sub)environments. Figures 4,5,6,7,8,9,10 and 11 show diverse macroscopic features, and Figs. 12 and 13 show diverse microscopic features of the different facies; the images in these figures are referred to Table 4.

Facies associations (FA) and depositional environments
The facies recognized in the studied Almonda valley outcrops are arranged vertically forming simple facies associations (FAs) or sequences that record the sedimentary processes that occurred in a particular sedimentation area over time. Four main FAs were recognized (Fig. 14A). These FAs represent cycles of variation in water flow conditions (hydrodynamics), water level and/or migration of active parts in the fluvial system (Arenas-Abad et al., 2010). The four FAs formed by deposition in active channel zones, either with free-flowing water (FAs 1 and 2) or with standing or sluggish flow in dammed areas (FAs 3, 4 and the uppermost part of 1). FA 4 could have also formed in the floodplain. These sequences can lack one or more features, resulting in several incomplete variants (e.g., sequence 1a in Fig. 14A).

FA 1: Barrage-cascade and dammed areas
This association represents the formation of a barrage-cascade from phytoclast accumulation and phytoherm growth and a dammed area upstream. It begins by erosion and then massive filling of a channel, initially with gravel and phytoclasts (facies G), and then evolves to dominant phytoclasts (Lph) (Fig. 5F, G). Locally, fine phytoclasts along with fragments of charophytes (Lphf and Lb) accumulated in areas that were protected from strong currents (Fig. 12D, E). The accumulation of phytoclasts (calcite coated stems of different sizes) could create an irregular channel floor, with areas acting as barrages. The progressive accumulation of moss layers (Lbr) in these areas would enlarge the barrages and produce cascades (Figs. 5B, E, 9B). In some areas, extensive microbial mats (Ls) alternated with moss layers (Figs. 5D, 9C, D, 10B), denoting changes in flow conditions. The faster The locations of the TNMC samples for AAR are indicated. B Moss and stromatolite boundstone forming a dome structure. On the right, note a smaller dome over a trunk mould lying horizontally (arrow). C Moss and stromatolite layers forming tight domes. The flow is towards the right (northeast). Note below a small cave with a stromatolitic speleothem coating. D Detail of stromatolite lamination in C. E Up-growing calcite-coated stem boundstone and associated phytoclasts, located between metres 4 and 5 in the MOICO section  Fig. 12F, G), as well as minor intraclasts and oncoids (Li and Lo). Charophyte stems and gyrogonites are preserved as fine particles in facies Sb and Lb (Fig. 12D). The shallow parts of these dammed areas could be colonized by hydrophilous plants, the submerged parts of which would be coated by calcite (facies Lst 1). The ACHOP build-up contains a good example of the complete development of this association (Fig. 5). Variants of this FA comprise (1) the occurrence of cascades without damming water upstream, as in the case of VSO build-up (Fig. 8A), in which stromatolite and laminated speleothem deposits are dominant (Figs. 8B, 8C, 13E); (2) the lack of coarse detrital sediments at the base, as in the MOICO buildup; and (3) the lack of barrage-cascade formation (variant 1a, Fig. 14A). The latter denotes a lower slope, which favoured the development of bedforms (i.e., cross-stratification within fine phytoclasts) and the development of extensive pooled areas, where diverse biota could live and organic matter-rich intervals could form. Variant 1a is present in RIBR and close to the RIB buildp-up (Fig. 7A, B).
This sequence is typical of steep stretches associated with dammed areas in stepped fluvial systems and has been recognized in many Quaternary fluvial systems, e.g., in central Spain, the Lagunas de Ruidera (Ordóñez et al., 2005) and western Turkey (Töker, 2017), and described in several other Quaternary examples of southern Italy (Violante et al., 1994)

FA 2: Low-slope channel
This association illustrates deposition in gently sloped channel areas where hydrophil-ous plants could live in shallow water, producing facies Lst 1. Increased energy episodes caused breakage of the already calcite-coated plants, and fragments accumulated in close areas (Lph, Li; Fig. 12A, B, C). Small domes could form from moss layers coating the phytoclast accumulations, pro- ducing facies Lbr. Water would be pooled upstream and/or between the domes; these small pools would be sites for the deposition of carbonate sand and silt (Sb), fine phytoclasts (Lphf) and rare intraclasts (Li). Aggradation would lead to the filling of these low-slope channel areas. This FA is common in the LAP build-up and the lower part of the MOICO build-up, e.g. metres 4-8 of the section (Fig. 9A, E). Similar facies associations have been described in Quaternary deposits of southern Italy (Violante et al., 1994) and Holocene deposits in the Añamaza valley, N of Spain (Arenas et al., 2014a), representing areas of low-energy water flow conditions along the valley. The lack of stromatolites in the domes would imply more homogenous flow conditions than in the case described in Spain.

FA 3: Shallow lakes and pools
Facies association 3 records deposition in shallow water areas of diverse extents. Typically, the most abundant facies were the bioclastic sands and silts (Sb) and less common intraclastic limestones (Li), which produced thick accumulations containing skeletons of ostracods and gastropods (Fig. 6B, D). Carbonate particles were derived from the breakage of upstream tufa and in situ calcite precipitation (lime mud) (Fig. 12F, G). Locally, some thin layers (i.e., 1-2 cm thick) contain microscopic organic matter, suggesting standing water conditions (Fig. 7B). Fine phytoclasts were introduced during slightly more agitated conditions, from the breakage of thin phytoclasts in littoral zones. These inputs were episodic and gave rise to cm-thick layering consisting of alternating Sb and Lphf facies, with less common millimetre-to 2-cm long oncoids (Lo; Fig. 7F). Filling and desiccation produced nodules and, in some cases, in situ intraclasts (breccia).  and Figs. 5,6,7,8,9,10,11 and 16 This association is well developed in the middle stretch of the Almonda valley, e.g., LAP,RIB and RIBR buildups (Figs. 6A,B,D,E,7A,B). Similar facies associations have been recognized in Quaternary deposits of southern Italy (Buccino et al., 1978), in Holocene deposits of the Lagunas de Ruidera (Ordóñez et al., 2005) and in Holocene deposits of the Añamaza valley (Arenas et al., 2014a), in the central and northeast Iberian Peninsula.

FA 4: Palustrine zones in dammed areas and floodplains
Facies association 4 was produced in flat zones of the valley with a dense cover of hydrophilous vegetation, including dammed areas along the channel and in the floodplain. The submerged parts of plants would be coated by calcite, producing extensive blankets of facies Lst 1 (Figs. 6C, 11A). Disruption by gentle water currents and accumulation of fragments in close areas would form facies Lph and minor oncoids (Lo). These vegetated areas could be flooded as a result of water flow increases during rainy seasons and/or aggradation of downstream barrages, as described in Holocene deposits of the Añamaza valley (Arenas et al., 2014a). Typical facies resulting from these flooding conditions are Sb, Lphf and Lb. Erosive concave-up surfaces and cross-stratification in facies Lphf denote periods of increased water inputs.
This FA is very common in the record studied herein. The RIB, LAP and MPAU buildp-ups show good examples of extensive palustrine zones (Figs. 6A, C, F, 11). Similar associations that formed in extensive palustrine zones are common in many low-slope river stretches and have been described in association with both fluvial and fluvial-lacustrine environments in the Mediterranean realm (Ajuaba et al., 2021;Arenas et al., 2014a;Henchiri, 2014;Pedley et al., 2003;Vázquez-Urbez et al., 2012) and in southern Portugal (Guerreiro, 2015).

Sedimentological characteristics of some build-ups
The geometry of the build-ups is not always evident due to the present vegetation cover and anthropic activities. In a few cases, wedge-shaped bodies consisting of moss boundstones (Lbr) and stromatolites (Ls) with increasing thickness downstream are observable, although these bodies represent partial development of larger structures (e.g., at VSO and ACHOP) (Figs. 5A, 8A).
It is worth highlighting the extensive areas occupied by some fine-grained carbonate and stem phytoherm deposits in the middle stretch of the Almonda River valley. These consist of a dominance of phytoclast rudstones and boundstones (Lph, Lphf, and Lst 1) and carbonate sands and silts (Sb). This is the case for the RIBR and RIB zones (Figs. 2, 6, 7) and many deposits in the Lapas locality (Figs. 2, 4E).

Upstream deposits
The ACHOP and RIBR buildups (13 and 22 m thick, respectively) contain gravel and conglomerate deposits at the base. The overlying carbonate deposits in ACHOP consist of alternating beds formed of moss boundstones and stromatolites (Lbr and Ls) and fine-grained carbonate deposits, i.e., phytoclast rudstones and boundstones (Lphf, Lst 1) and carbonate sand and silt (Sb) (Fig. 5), conforming to FA 1. The overlying carbonate deposits in the RIBR build-up contain stacked, decimetrethick and metre to decametre-long lenticular bodies formed of facies Lphf and Sb, at places displaying cross-stratification (Figs. 4A, 7D), as expressed by FA 1a and 4.

Moinho do Pau deposits
This build-up reaches 23 m thick (Fig. 2 show the location). The outcropping deposits are rather tabular. At the base, these deposits consist of sands and silts (Sb), either massive or banded, which are overlaid by dominant phytoclast rudstones (Lph) and laterally continuous up-growing stem boundstones (Ls 1) (Figs. 4C, 11). These deposits formed via lime mud and sand deposition in wide dammed areas, which later became colonized by dense covers of hydrophilous plants, the submerged parts of which were coated by calcite. This palustrine zone was subject to low-energy flooding processes and/or crossed by draining channels, which caused the breaking of already calcite-coated stems; the fragments accumulated without much transport, almost on-site. Depositional conditions for this build-up are represented by FA 4.

Lapas deposits
This extensive build-up that reaches 16 m thick, stands out because of the tabular to lenticular geometry consisting mainly of an abundance of fine-grained carbonate deposits (i.e., Sb, Lphf), which are at places banded and laminated and showing depositional inclination. These deposits are associated with phytoclast rudstones and stem boundstones (Lph, Lst 1; Fig. 7E) and less common 20-to 50 cm thick layers consisting of small oncoids (Lo; Fig. 7F). At places, horizontal-lying thin-stem bunches that appear visually similar to charophytes mark the flow directions (Fig. 7E). These deposits formed in large dammed areas, with slow flowing water and bearing palustrine vegetation (e.g., as represented by FA 3 and 4). Moreover, at the downstream end of the Lapas build-up, highly inclined moss and stromatolite layers (Lbr and Ls) and down-growing calcite-coated stems (Lst 2), that together form hemidomic bodies (Fig. 8D, E), reflect deposition in a small cascade that could have dammed water upstream, allowing the thick and extensive deposition of fine-grained deposits (i.e., conforming to FA 1 or 2).

Moinho da Cova deposits
The lowermost outcropping portion consists of phytoclastic rudstones and boundstones (Lph and Lst 1; Fig. 9E) that constitute extensive 1 3 tabular layers, associated with small domes formed of bryophyte boundstones (Lbr), which produce inclined lenticular beds, i.e., as in FA 2. This portion of the build-up represents deposition in flat areas with abundant palustrine zones. On the southern bank of the present river, equivalent layers are dominated by phytoclast and intraclast tabular beds (Lph and Li) (Fig. 10F), indicating dominant breakage processes by currents.
The upper outcropping portion is dominated by thick and extensive bryophyte boundstones and stromatolites (Lbr and Ls), most of which form highly inclined lenticular or prismshaped beds, with interbedded lenses consisting of calcitecoated hanging plants, Lst 2 (Fig. 9B, C). These facies corresponded to cascades or barrage-cascades structures (e.g., as in FA 1) that were at least of 2 m high and would grow across an extensive area of the river. This was a steep part of the valley that was broad enough to allow wide cascades and variations in the main channel direction. The main water direction in this deposit was northwards.
An erosional surface separates these phytohermal bodies from the uppermost deposits, which are up to 1.5 m thick. These deposits consist of phytoclast and intraclast limestones and carbonate sand patches that together form a discontinuous and rather chaotic deposit ( Fig. 4F; 10E). These characteristics reflect deposition after a high-energy flow episode, which would have eroded the previous deposits, as reflected by the lower half of FA 1a.

Stable isotope composition
The studied tufaδ 13 C values range from − 7.97 to − 10.84 ‰ V-PDB, and the δ 18 O values range from − 4.07 to − 5.27 ‰ V-PDB, with mean values of − 10.11 and − 4.54 ‰ V-PDB, respectively (Table 4). These values are typical of meteoric fresh water with 12 CO 2 influence (e.g., Andrews, 2006;Deocampo, 2010;Faure, 1998;Leng & Marshall, 2004). Differences in δ 13 C between the different facies are small, except for four samples corresponding to facies Lst 2, Ls and Lbr (Lsp), two in each of the two analysed sections, which yield less negative values. Differences in δ 18 O between the different facies are smaller. In terms of the section, mean δ 13 C and δ 18 O values of all samples are more negative in the MOICO section than in the ACHOP section (Fig. 15). Figure 16 presents a sedimentary facies model for the study area. The area is noticeable by the extensive development of facies formed in still and slow-flowing water in dammed areas located along the middle stretch of the Almonda valley, coinciding with the lowest river slope stretch at present. These dammed areas provided sites for the deposition of fine carbonate sediment (Sb, Lb and Lphf), small oncolite formation (Lo) and growth of hydrophilous plants whose submerged parts became coated by calcite in shallow areas (Lst 1). Breakage of the latter would produce facies Lph (FAs 3 and 4). Gently sloped channel stretches were also present between knickpoints, favouring the development of vegetated zones (Lst 1) and the formation of phytoclasts (Lph), small oncoids (Lo) and small mounds consisting of moss mats (Lbr), as reflected by FA 2.

Proposal of a sedimentary facies model
Facies denoting the presence of knickpoints typically form highly to moderately inclined beds, i.e., moss boundstones and stromatolites that developed in cascades and barrage-cascades. Calcification of moss mats and cyanobacterial mats that grew on step surfaces under moderate-to fast-flowing water produced mounds formed by Lbr and Ls (FA 1). Progradation of these structures caused the formation of small caves below and behind the cascade front, with speleothem deposits. These facies occur in a few places in the Almonda valley, e.g. in the ACHOP build-up (Fig. 5E), at the downstream end of the Lapas build-up (e.g., Fig. 8D, E)) and at Torres Novas, e.g. the MOICO build-up (Fig. 3). The visible height of these structures reaches 3 m.
Presently, the overall slope of the Almonda River is very low (0.33%), but has several areas with higher slopes (Fig. 3). Immediately after the source, along 270 m, the  Table 3 for values slope is steep (4%). After this point, the river floor shows several sections with an almost zero slope (0.00-0.03%), interspersed with sections with a slope of less than 1.5%, and the two steepest sections, with a 5.5% slope between Lapas and Torres Novas and a 4.6% slope in the Torres Novas locality (Fig. 3). The slope of the Quaternary Almonda valley was also small, as inferred from the position of the base of the tufa build-ups along the valley, the short height of the cascade-related structures and the large extent of deposits formed in dammed areas, which denotes a large separation between consecutive knickpoints. Therefore, the Quaternary tufas and allied sediments formed in a low-to moderatelysloped fluvial valley with a single watercourse and presented several short knickpoints and extensive flat areas between them. Local lateral tributaries were present in the upstream reach. This low-to moderately-sloped stepped fluvial system consisted of cascades, barrage-cascades and large slowflowing and standing water areas. Palustrine conditions were present in the latter and in pools on the floodplain. Incision by channels occurred before or at the beginning of tufa deposition, as reflected by gravel and conglomerate deposits at the base of the several tufa build-ups (e.g., ACHOP and RIBR build-ups). Rare detrital deposits through some tufa sections denote later, much less intense and infrequent erosional processes. Overall, these erosional processes were related to episodes of increased precipitation that affected the Mesozoic and Cenozoic bedrock, causing higher water flow and providing the Pleistocene Almonda streams with gravels and intraclasts.
Despite the lack of clean outcrops along the flow direction, the carbonate facies distribution along the valley corresponds to a depositional architecture consisting of wedgeshaped bodies that thicken downstream (cf. Arenas et al., 2014a). Within each wedge-shaped body, the larger extent corresponds to facies Sb, Lph, Lphf, Lst 1 and Lo, which formed in low slope areas. In contrast, the downstream portion of the wedge-shaped bodies represents short spaces having facies that formed in moderate-to high-slope substrates, such as Ls, Lbr and Lst 2, which together produced the thickest deposits within a single wedge-shaped body. This depositional arrangement is not different from the sedimentary facies models described for other stepped fluvial carbonate systems (Arenas et al., 2014a;Pedley, 2009;Vázquez-Urbez et al., 2012;Violante et al., 1994). Figure 16 presents a sketch that summarizes the most important features of the Pleistocene Almonda fluvial system. These features are as follows: 1) The first upstream tufa deposits occur approximately 2.5 km downstream of the Jurassic rocks; the main springs at present stem from the Jurassic rocks. This tufa occurrence coincides with the first upstream change in slope of the present river. 2) Gravel deposits occur mostly at the base of some tufa build-ups; these deposits formed in relation to high discharge events that provoked the erosion of previous rocks, incision into the Cenozoic bedrock and then deposition of polymict gravel (see Table 4). 3) Extensive deposits that formed in still and slow-flowing water areas occur in the middle stretch, between kilometres 5 and 8 of the valley, which at present corresponds to a very low slope area. 4) Palustrine conditions were extensive in association with large flat dammed areas and floodplains, at locations affected by currents that caused breaking of calcitecoated plants, with the resulting fragments accumulating very close to their origin. 5) Stromatolites occur in association with moss boundstones in moderate-to high-slope structures (stepped and vertical cascades), which represent short knickpoints along the river course. 6) The thickest outcropping deposits are in the downstream stretch, coinciding with relatively larger changes in slope.

Significance of the stable isotopic composition
The stable isotope values of the analysed samples are within the range of other fluvial tufas (mean δ 13 C = − 10.11‰ V-PDB; mean δ 18 O = − 4.54‰ V-PDB). The δ 13 C values are typical of freshwater carbonates, and they are closer to the more negative extreme. The δ 18 O values are biased towards the less negative extreme (Andrews, 2006;Faure, 1998). Isotopic analyses of diverse groundwater points in the Almonda hydrographic basin yielded values between − 3.5 and − 5.3 ‰ for δ 18 O and − 18.7 and − 30.1 for δ 2 H, with mean values of − 4.6 and − 24.5 ‰ V-SMOW (Antunes, 2016). Plotting of these values yielded a line that is close to the global meteoric line, thus indicating that the main recharge of the aquifer is precipitation, with negligible effects of altitude or continentality (Antunes, 2016).
The available δ 18 O analyses of precipitation (rainfall) at several stations across Portugal (IAEA, 2005; Global Network of Isotopes in Precipitation-GNIP; https:// www. iaea. org/ servi ces/ netwo rks/ gnip; Carreira et al., 2009) can tentatively be compared with the groundwater δ 18 O in the Almonda hydrographic basin. The altitude (maximum is 679 m; mean is 380 m) and mean precipitation (1200-1400 mm/year; https:// www. ipma. pt/ pt/ educa tiva/ tempo. clima/ index. jsp? page= clima. pt. xml) of Serra de Aire are similar to those of Portalegre and Vila Real stations, with δ 18 O values of − 5.6 and -6 to 4 ‰ V-SMOW, respectively. The values of Portalegre and Vila Real stations are more negative than the − 3.5 to − 5.3 ‰ range of the groundwater in the Almonda basin. Assuming similar or slightly less negative δ 18 O values for precipitation water in the Serra de Aire than in the Portalegre and Vila Real stations (due to its closer position to the coast), the difference may partially be due to the effects of residence time of water in the aquifer before water outflowing. A comparison with the Porto station would not be appropriate given the coastal position and low altitude, with δ 18 O = − 4.54 ‰ V-SMOW.
The δ 18 O values of the Almonda tufas (mean = − 4.54‰ V-PDB, N = 14) are coherent with values of fluvial tufas formed in a temperate-climate zone, where there is sufficient throughflow for evaporation and residence time effects on river water to be considered negligible (cf. Andrews et al., 1997). Accordingly, the relatively high tufa δ 18 O values may reflect an already 18 O-enriched river water, reflecting an initially 18 O-enriched water precipitation that is consistent with the proximity to the evaporation source and likely affected by the abovementioned residence time in the aquifer. Pertinent processes that may have amplified such enrichment in the fluvial environment are 16 O loss by flow agitation in turbulent areas, and fractionation effects due to temperature and calcite precipitation (Ortiz et al., 2006;Osácar et al., 2013).
The tufa δ 18 O values in the Almonda valley are less negative than the values obtained in other fluvial deposits in central and NE Spain, both for Quaternary and recent tufas. For instance, in the Tagus River Basin, central Spain, Quaternary tufa deposits show mean δ 18 O values of -6.72‰ V-PDB, N = 74 (Ortiz et al., 2009). In the Piedra River valley, NE Iberian Peninsula, the Quaternary mean δ 18 O = − 7.62‰ V-PDB, N = 62 (Vázquez-Urbez et al., 2011), and recent tufa δ 18 O values ranged from − 7.13 to − 9.28‰ V-PDB (Osácar et al., 2013). The recent water δ 18 O values are less negative in the Almonda basin (− 4.6 ‰ V-SMOW; Antunes, 2016) than in the Piedra River (-8.5 ‰ V-SMOW; Osácar et al., 2013). The latter difference may reflect the different 18 O depletion from the evaporation source to inland areas (cf. Faure, 1998).
Typically, the δ 13 C values of the studied tufa (mean δ 13 C = − 10.11‰ V-PDB, N = 14) reflect the input of soilderived and/or organic matter-derived CO 2 in the water (Chafetz et al., 1991;Leng & Marshall, 2004). Two external sources of carbon could contribute to lowering the δ 13 C of dissolved inorganic carbon (DIC) of water from which calcium carbonate would have precipitated: one is the dense vegetation cover on the catchment rocks and the other is the lithological composition of the aquifer rock (cf., Faure, 1998). It is worth highlighting the presence of bituminous shales in the Upper Jurassic sequence -described as "fetid bituminous marly limestones" by Manuppella et al., (2000, p. 65)-of the Estremenho Massif (Serra de Aire), which constitutes the catchment area. A long residence time of water in an aquifer can favour water enrichment in 12 C from both soil-derived and bituminous organic matter-derived CO 2 . This 12 C signature would greatly counteract the 13 C enrichment effects due to water being in contact with the aquifer marine carbonates.
The δ 13 C values of the Almonda tufas are more negative than most Quaternary tufas in central and NE Spain. For instance, in the Tagus River Basin, central Spain, tufas show a mean δ 13 C values of − 7.10‰ V-PDB, N = 74 (Ortiz et al., 2009), and in the NE of the Iberian Peninsula, the Piedra valley tufas have a mean δ 13 C = − 6.88‰ V-PDB (N = 62), and the Añamaza valley tufas have a mean δ 13 C = − 6.01‰ V-PDB (N = 65) (Vázquez-Urbez et al., 2011). This difference might be related to a series of factors concerning soilderived CO 2 , as tufa δ 13 C values lower than − 8‰ V-PDB usually indicate strong soil zone influence (Andrews et al., 1997). These factors include the characteristics of soils (e.g., thickness and type of soil vegetation) and the catchment rock composition and other basin-related features.
Apart from the production of plant-and soil-derived CO 2 within the tufa system as a result of the abundant plant cover, a series of intrinsic processes that influence the δ 13 C of the DIC of fluvial water includes (Andrews, 2006;Pentecost & Spiro, 1990): (1) the preferential 12 CO uptake by plants; (2) the 12 CO 2 output due to degassing, mainly in rapids and cascades, and (3) reaching an equilibrium between atmospheric and water CO 2 . The three involve increasing δ 13 C values in water DIC. The isotopic fractionation produces calcite δ 13 C that is less negative than the water δ 13 C from which it was precipitated (Faure, 1998;Osácar et al., 2013). Assessing the importance of each process and its imprint on calcite δ 13 C is not a straightforward approach and usually involves complex experimental approaches (Arp et al., 2001(Arp et al., , 2010.
In brief, water in the aquifer would have provided the outflowing Almonda River water with 12 C-rich and 18 O-rich CO 2 , with the effects of further physical and biological processes of less importance on the stable isotope composition of tufa.
For the slightly more negative mean stable isotope values of MOICO compared to ACHOP samples, a speleothem sample in ACHOP greatly contributes to increasing the mean value. Except for that sample, the remaining values are similar. The increasing δ 13 C values of the MOICO samples over time may reflect a decreasing input of 12 C in relation to the lesser development of soils, perhaps as a result of decreasing precipitation. These ideas must be taken as tentative given that a larger number of analysed samples would allow more conclusive results.

Factors influencing tufa formation
A wide array of factors can intervene in fluvial carbonate formation Arenas-Abad et al., 2010;Della Porta, 2015;Arenas et al., 2014a;Capezzuoli et al., 2014;Toker, 2017;among others). Extrinsic factors include climate, tectonics, and bedrock composition and structure, including the characteristics of the aquifer feeding the fluvial systems. Intrinsic factors correspond to the presence of vegetation, physical and chemical properties of water and CO 2 degassing (Arenas-Abad et al., 2010). Moreover, some of these factors, i.e., temperature, may influence tufa formation on different scales, imprinting both cyclic and noncyclic climate signatures on the deposits.
The water composition of the Quaternary Almonda River had to be saturated in Ca +2 and HCO 3 − to form tufa calcite, a requirement that has been shown in studies of recent tufa-forming streams (Auqué et al., 2014;Kano et al., 2007;Kawai et al., 2006). Water coming from the Jurassic rock aquifer in the Serra de Aire provided such a required composition. The water composition of the present-day Almonda River supports such a statement (Antunes, 2016).
Moreover, as has been observed in recent streams, a certain distance from the spring is usually required for water pCO 2 loss to equilibrate with the pCO 2 atmosphere; then, pH increases and allows calcium carbonate to precipitate (Auqué et al., 2014;Chen et al., 2004;Liu et al., 1995). Calcite saturation levels increase downstream coinciding with the increase in the river slope, which enhances CO 2 loss, thus promoting rapid physico-chemical precipitation of calcite (Chen et al., 2004;Drysdale et al., 2002). In the Almonda case, the first tufa deposits along the main course occur approximately 2.5 km away from the present spring at the Jurassic/Cenozoic contact, which is a much shorter distance than in the Pleistocene tufa of the Piedra River (Arenas et al., 2014b).
Variations in water discharge can also influence the distance between springs and first tufa deposits (Arenas et al., 2014a;Drysdale et al., 2002). Accordingly, the comparatively short distance between tufa deposits and the Jurassic/ Cenozoic contact might be indicative of moderate water discharge during deposition of the Pleistocene Almonda tufas. Nonetheless, studies of other tufas in nearby valleys would help discern this proposal.

Controls on the depositional architecture
Many present-day fluvial tufas are formed in streams that flow along valleys with variable slopes (Ford & Pedley, 1996;Pentecost, 2005). Tufas forming along stepped longitudinal profiles are the most common. In some cases, tufas develop as perched spring-line deposits on the valley sides and in very low slope areas within palustrine contexts (Arenas-Abad et al., 2010;Henchiri, 2014;Özkul et al., 2010;Pedley, 1990;Pedley et al., 2003). The studied tufa and allied deposits of the Almonda River valley formed in a stepped fluvial valley with a low to moderate slope.
Collectively, the different studies show that the geometry of the sedimentary fill mostly depends on extrinsic factors that condition the topography of the basin floor, such as changes in bedrock composition and structure. Peña et al. (2000) and Toker (2017) described examples in which the presence of faults conditioned the floor topography and depositional fill geometry.
For the most part, abrupt changes in slope along the valley are related to changes in bedrock lithology, which offers different resistance to erosion (Arenas et al., 2014a;Vázquez-Urbez et al., 2012). These initial knickpoints favour the creation of cascades and rapids, on which tufa development is faster, due to the greater CO 2 loss in fast and turbulent flowing conditions (Violante et al., 1994;Pentecost, 2005;Kawai et al., 2006;Kano et al., 2007;Gradziński, 2010). Some areas in which trunks and calcite-coated plant debris accumulate are prone to produce cascades or barrage-cascades that pool water upstream. Faster deposition in all these areas with fast-flowing water tends to enlarge the initial knickpoints, resulting in the formation of a stepped longitudinal section with rapids, cascades, barrage-cascades, and still and slow-flowing dammed water areas (i.e., pools or lakes) (Arenas et al., 2014a).
As described above, the depositional geometry of the Pleistocene Almonda deposits corresponds to large, laterally related wedge-shaped bodies that thicken downstream, i.e., wedges with a large length/height (L/H) ratio formed in stretches with gentle slopes. In each wedge, a significant extent of deposits corresponds to the dammed facies and the other gentle-slope facies (i.e., fine-grained sediment, small oncoids, phytoclasts and up-growing calcite-coated stems). A much smaller extent corresponds to the high slope facies formed in moderate-to fast-flowing water (i.e., moss boundstones, stromatolites and down-growing stem boundstones).
The lack of large or high knickpoints and overall small to moderate slope of the Pleistocene Almonda River, with extensive gently sloped to flat areas corresponding to a large length/height (L/H) ratio, are likely related to 1) the absence of abrupt changes in the lithological composition of the bedrock, which is soft marls and limestones, and these rocks easily erodible, and 2) the bedrock structure, with a very small dip of the strata that the river crossed. The path of the Pleistocene Almonda River had almost perpendicular segments to the strike of the strata, which favoured small changes in slope when crossing rocks of different resistance. However, this situation was very different from other examples in which the bedrock consisted of lithological units with more contrasted resistance to erosion along the river path and with higher dip, causing higher knickpoints. This is the case for several Quaternary fluvial tufas in northeastern Spain (Arenas et al., 2014a;Vázquez-Urbez et al., 2012). Together these results show the strong control of bedrock composition and structure on the depositional architecture of fluvial carbonate deposits.
These results are consistent with other tufa studies. For instance, in the Pleistocene Ebrón River valley, NE of the Iberian Peninsula, the existence of a low sloped downstream stretch favoured the development of a dominant carbonate palustrine system (Ajuaba et al., 2021). There, the type of tufa facies and the depositional architecture clearly reflect the increase in width and decrease in slope of the sedimentary area as a result of the bedrock lithological and structural changes, from more resistant Mesozoic rocks upstream to less resistant Neogene detrital rocks with smaller dips downstream, together allowing a greater width and smaller slope of the sedimentary area downstream.

Climate and hydrology
Most tufa and allied deposits in the studied area herein appear to have formed during MIS 5 (likely MIS 5e and 5c; Gibbard et al., 2005). This stage was characterized by warm temperatures, which favoured the formation of important deposits of tufa in many high-and mid-latitude areas throughout the world, which are very abundant in the Mediterranean region (Durán, 1989;Martín Algarra et al., 2003;Pedley, 2009;Sancho et al., 2015). Precipitation seems to be a limiting factor in low-latitude regions and is more dependent on local to regional atmospheric conditions . The sensitivity of the Iberian Peninsula to shifts in the North Atlantic atmospheric belt makes any precipitation model more complex (Cacho et al., 2002). However, the proximity of the study area to the Atlantic humidity could favour the continuous recharge of the Mesozoic rock aquifer, thus assuring water availability during the formation of the studied tufas.
The outcropping studied record does not show evidence of vigorous erosion phases, except for the upper part of the MOICO build-up. This fact might be related to water flow regimes of little variation in intensity (i.e., with stable discharge), a characteristic that may be indicative of stable precipitation conditions, as occurs in temperate humid climate zones. Some studies have shown that highly variable flow regimes in arid and semiarid conditions can cause erosion phases and deposition of coarse clastic sediments that alternate with carbonate deposits , while rare floodrelated deposits have been reported from tropical conditions in northern Australia (Carthew et al., 2003). Together, these facts suggest that the deposition of the Pleistocene Almonda tufas might have occurred under quite stable precipitation regimes, which allowed continuous water flow through the river and abundant hydrophilous plants to live in still and slow-flowing water zones.

Scientific interest and potential of tufas as natural, cultural and economic resources
This part of the text aims to highlight the importance of the use and preservation of tufas as natural, cultural and economic resources, considering the diverse scenarios in which tufas can stand out in a region: rock quarrying and building, water uses in agriculture, industry and urban areas, landscape, tourism, and cultural-educational activities. In addition to the vast scientific interest, the Almonda River valley is an extraordinary example to illustrate the management of tufa-depositing streams and their deposits as natural, cultural and economic resources in a region. At present, some industrial and agricultural activities in the region are sourced with water derived from different points along the river, but primarily upstream. In the past, water mills ("moinhos") and dams were built in zones with slope changes. These anthropogenic elements indicate the type of tufa deposits and structures that developed along the river course in the past.
Some urban zones are built on tufa deposits, but tufa observation is possible, at least partially, from natural sections and open quarrying and long-cave digging. Paradoxically, the anthropogenic activity related to open-air tufa quarrying for building uses has provided numerous new outcrops that have broadened the available areas for diverse geological and paleontological studies. Several large, ancient open-air quarries are located in Moinho da Cova-Torres Novas, Alto do Choupal-Ribeira Ruiva, and Moinho do Pau-Lapas, some of which are more than 100 m long. There are a number of smaller quarries throughout the valley. Underground extraction of tufa has been accomplished by means of digging tunnels, which offer an opportunity to study the fluvial carbonate deposits of the Almonda River. Among the several examples, the extraordinarily complex net of tunnels excavated below the Lapas town that occupy most of the fluvial terrace stands out, spanning approximately 700 m of length, of which more than 400 m are opened to visitors (Grutas de Lapas). The age and origin of these tunnels are still unclear. Zbyszewski et al. (1974) referred to the existence of pre-historical caves in Ribeira Branca and Lapas. Santos and Reis (2021) considered the oldest items found in the tunnels could be atributed to a period between the Mesolithic and the Chalcolithic, within 8000-3000 years AC. However, the authors of the present contribution estimate that some natural, fluviokarstic tunnels may have formed during MIS 2-4.
Tufa extraction for building uses dates back to the medieval and ancient epochs, denoting the close relation between the geological resources and the cultural heritage of the region. This is the case for the Roman village Cardilium in Torres Novas. Many examples of public and private buildings using tufa as building material are also found throughout the Mediterranean zone.
As in the case of other tufa-depositing streams, the landscape along the Almonda River valley is spectacular, representing a touristic attraction. There are a number of paths with explanatory panels at significant points along the present river course. The sharp transition from the Estremenho massif to the Tagus Basin at Arrifes shows the karstic spring that feeds the Almonda River, stemming from the largest groundwater system in Portugal (Gruta do Almonda). This is a touristic attraction as well. One important and still pending point is the study of the present-day sedimentation process. The results can provide crucial information on tufa facies, deposition rates and hydrochemical composition, which can be used in future actions in the region and compared with other cases (e.g., the Piedra and Mesa Rivers in the NE Iberian Peninsula, and the Korana and Krka Rivers, in Croatia).
A great push toward tourism in the Almonda River region was achieved because of the geo-touristic requalification of the Grutas de Lapas by the Torres Novas municipality in 2018. A visitor centre with diverse information (geology, geography, history and archaeology) explains the varied tufa-related heritage in the region, underlying the geological significance (Fernandes, 2018). That action was awarded by ProGeo-Portugal.
The geological information offered in this paper hopes to raise the scientific interest in the tufas of this Portuguese region and elsewhere as tools for understanding the environmental, climatic and hydrological conditions of past scenarios. Moreover, the example of the Almonda valley tufas serves to emphasize that the correct management of present and ancient tufa-related contexts is crucial to value and preserve the varied tufa-related heritage.

Conclusions
The tufas and associated deposits of the Almonda River valley were deposited during the Middle-Late Pleistocene (MIS 5) in a low-to moderately sloped stepped fluvial system. It consisted of short barrages and barrage-cascades separated by large areas occupied by slow-flowing and still water. The latter were sites for diverse aquatic fauna and flora to live and for fine-grained sediment (carbonate sand, lime mud, fine phytoclasts, small oncoids and intraclasts) and up-growing stem boundstones to form. The moderate-to high-slope areas were sites for stromatolite, moss and down-growing stem boundstones to form. Incision by channels occurred before or at the beginning of tufa deposition. The homogeneous bedrock lithology and structure propitiated the small to moderate riverbed slope, which was later exaggerated by ongoing tufa growth at knickpoints.
Mostly stable precipitation regimes during the study tufa deposition are deduced from the lack of frequent and/ or intense erosional phases and the preservation of thick fine-grained sediments and palustrine facies.
Water springs stemming from the Jurassic rock aquifer provided water with sufficient Ca ++ and HCO − 3 to reach calcite saturation levels a short distance downstream. The calcite δ 13 C values of tufa facies reflect the influence of 12 C-rich CO 2 sourced by bituminous rocks and vegetation cover in the catchment area. The calcite δ 18 O values suggest initially 18 O-rich CO 2 in the aquifer water, which was likely also affected by 18 O-enrichment due to residence time. The proximity of the study area to the Atlantic humid zone caused precipitation to be 16 O-depleted. Differences in calcite δ 18 O composition with respect to Quaternary tufas in the central and eastern Iberian Peninsula support the differential 18 O depletion from the evaporation source to inland areas. The proximity also favoured the water recharge of the Mesozoic rock aquifer, assuring water availability during the formation of the studied tufas.
Open access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long … Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.