Palaeoenvironmental setting of lacustrine stromatolites in the Miocene Wudaoliang Group, northern Tibetan Plateau

Abstract

Lacustrine stromatolites were widespread in the Miocene Wudaoliang Group (stromatolites of the Wudaoliang Group), northern Tibetan Plateau; but only at one location nearby the Wudaoliang Town, they occurred intensively in thick, laterally traceable beds (Wudaoliang stromatolites). Although deposited in lacustrine environment, the lack of fossils in these rocks hampers determining whether the stromatolites formed in freshwater or saline conditions. To address this problem, and in an attempt to identify criteria to distinguish differences of freshwater and saline conditions, we studied the laminae microfabrics, stable carbon and oxygen isotope ratios, rare earth element patterns and biomarkers of the stromatolites. These stromatolites can be divided into fenestral stromatolites and agglutinated stromatolites. The fabric of fenestral stromatolites is formed by microcrystalline carbonate enclosing spar-cemented, angular crystal traces. Essentially, this fabric is interpreted as pseudomorph after former formed evaporite crystals. Faecal pellets identical to that of the present-day brine shrimp Artemia, lack of other eukaryotic fossils, and stable isotopic signals point to a shallow, evaporation-dominated hypersaline lake setting. Covariation of carbon and oxygen isotopes indicates hydrologically closed conditions of the Miocene lake on northern Tibetan Plateau.

However, if compared to other lacustrine carbonates of the Wudaoliang Group, the high δ¹³C values of the investigated Wudaoliang stromatolites reveal an additional photosynthetic effect during the deposition of the stromatolites. Furthermore, although no direct evidence is available from field observations and microfabrics, a positive europium anomaly of Wudaoliang stromatolites indicates that a palaeo-hydrothermal inflow system had existed in the outcrop area. These new results favour a hypersaline lake setting subject to hot spring inflow for the Wudaoliang stromatolites, in contrast to earlier interpretations suggesting a freshwater lake setting (e.g. Yi et al., Journal of Mineralogy and Petrology 28: 106–113, 2008; Zeng et al., Journal of Mineralogy and Petrology 31: 111–119, 2011). This approach may be appropriate for other lacustrine, unfossiliferous microbialites in settings where the environmental conditions are difficult to determine.

1 Introduction

Microfabrics and geochemical features of lacustrine stromatolites are valuable indicators of palaeoenvironmental changes in continental settings otherwise rare in fossils (Monty 1976; Surdam and Wray 1976; Riding 2000). In this study, the environmental conditions of extensive lacustrine stromatolites are investigated using a combination of petrography and geochemistry to resolve existing issues about controls on their formation. Lacustrine stromatolites are common in the Miocene Wudaoliang Group in the northern Tibetan Plateau region (Qinghai Province), western China, and they usually occur as centimeter- to decimeter-thick beds intercalated between carbonates. But nearby Wudaoliang Town in particular, the stromatolites (also from the Wudaoliang Group) occur in an area of about 240 × 80 m² in size, forming a total 15-m-thick bed, exposed on a hill slope (Fig. 1). We refer the former, centimeter- to decimeter-thick beds as stromatolites in the Wudaoliang Group and the latter nearby Wudaoliang Town, specifically, as Wudaoliang stromatolites. Yi et al. (2008) and Zeng et al. (2011) inferred from stable isotopes that the stromatolites in the Wudaoliang Group formed in a freshwater lake and they interpreted a humid interval of the long-term, warm and dry Miocene palaeoclimate in northern Tibetan Plateau. However, typical skeletal fossils are very rare in stromatolites in the Wudaoliang Group, in contrast to the common occurrence of calcified cyanobacteria (Riding 1991), gastropods and ostracods (e.g., the Eocene Green River Formation, Surdam and Wray 1976) in freshwater lakes or rivers. Wudaoliang stromatolites possess substantial elongated, filament-like microcrystalline structures, with angular spar-cemented voids in between, reminiscent of former evaporite crystals. Similar crystal traces are known from other calcareous stromatolites of hypersaline settings (e.g., Arp et al. 2008). Furthermore, syn-sedimentary Pb-Zn deposits in the adjacent Tuotuohe Basin (Hao et al. 2015) suggest a possible hydrothermal impact also in the Wudaoliang Basin at that time (Zhang et al. 2015).

Fig. 1

a Stromatolite-bearing core profile of Zk1 drill in the Miocene Wudaoliang Group (modified from Wu et al. 2009 and Zeng et al. 2011) and geographic location of the Wudaoliang sampling site (Inset map of China is modified after State Bureau of Surveying and Mapping, GS (2016)1591). Distance between the Zk1 drill site and the sample locality (93°05′15.8″E, 35°13′21.0″N) is about 2 km. Note the stromatolite intercalated beds are only in 10~50 cm thick in Zk1 drill core; b Satellite image shows the stromatolite outcrop (about 10 m in thickness) nearby Wudaoliang Town. The box area is approximately 240 × 80 m² in size; c Field photo shows the stromatolite outcrop from the view of another hill

Thus, it is unknown whether these stromatolites formed in a freshwater setting, saline halite lake or soda lake conditions. Meanwhile, a possible hydrothermal impact on these stromatolites remains to be confirmed. In order to solve the undetermined issues of environmental conditions, this study examined the macrofabrics and microfabrics of the stromatolites, distinguished different laminae by the fabrics and geochemical composition, and discussed the genesis of the stromatolites. LA-ICP-MS was used to obtain high-resolution elemental data to analyse the potentially preferential elemental enrichment in different laminae and fabrics. Combined these results with stable oxygen and carbon isotopes, consequently, this study identified one kind of hypersaline lake environment during the Miocene stromatolites formation period, nearby the present Wudaoliang Town of Qinghai Province, northern Tibetan Plateau.

2 Geological settings

The investigated stromatolites belong to the carbonate-dominated Miocene Wudaoliang Group (marls, dolomites and limestones; 23.5–13.5 Ma) that unconformably overlies the Oligocene conglomerates and sandstones (Yaxicuo and Fenghuoshan Groups) in Wudaoliang and adjacent basins (Wu et al. 2008) in Hoh Xil area. Wudaoliang Group, Yaxicuo Group and Fenghuoshan Group were deposited in a lacustrine environment that developed in the northern Tibetan Plateau during its uplift, crustal shortening, and a general three-phase evolution of warm (dry)-cool (wet)-cool (dry) climate during the Oligocene-Miocene (Wu et al. 2008; Wang et al. 2008). The interconnected early-Miocene lake-basins were successively separated during their later evolution.

The 310–350-m-thick Wudaoliang Group shows only minor tectonic deformation, with inclined beds of tipping angle less than 30°, suggesting a comparatively weak tectonic-active episode in the Cenozoic of the northern Tibetan Plateau (Wang et al. 2002, 2008). The lower part of the Wudaoliang Group hosted marls with a wide spectrum of fresh- to brackish-water organisms (ostracods, gastropods), while its upper part was dominated by dolomites and limestones with intercalated stromatolites (Wu et al. 2008; Zeng et al. 2011).

The sampling site was located nearby Wudaoliang Town along the Qinghai-Tibet Highway (G109), 270 km south from the Golmud (GPS: 93°05′15.8″E, 35°13′21.0″N, H 4667 m; Fig. 1).

3 Material and methods

Three large thin-sections (all with a thickness of 80 μm, and two with a size of 10 × 15 cm² and another of 4.5 × 1.5 cm²) and twenty standard thin-sections (30 μm thick) were used for petrographic investigation on a Zeiss Axioplan microscope (Carl Zeiss MicroImaging GmbH, Jena, Germany). Micro features were carefully studied under the microscope, to develop a clear understanding of the sedimentary fabric, before geochemical approaches were used.

3.1 Stable oxygen and carbon isotope analysis

One fresh stromatolite cut slab was drilled to obtain microcrystalline calcite samples from each individual lamina. In each lamina at least two microcrystalline calcite samples were drilled. Totally, 89 calcite samples were obtained in 36 continuous, alternating dense and porous laminae towards one linear direction within one stromatolite cut slab. The 89 powdered calcite samples were then prepared and analysed by a Finnigan MAT 252 mass spectrometer. Both carbon and oxygen isotopes of the stromatolitic limestone were analyzed under the Vienna Peedee Belemnite standard (VPDB) and the results were given in δ‰.

3.2 LA-ICP-MS analysis

A 193 nm ArF excimer laser ablation system (LA, GeoLas, COMPEX 110, Lambda Physix, Germany) coupled to an Elan DRCII 6100 inductively coupled plasma-mass spectrometer (ICP-MS, PerkinElmer, Germany) was used for rare earth elemental measurements. The laser ablation was operated at a wavelength of 193 nm, energy density of 10 J·cm^− 2 and frequency of 10 Hz. Ablation spot size was set to 50 μm, and one fenestral stromatolite sample was analysed by the ablation spot mode: 120 calcite samples in 17 laminae were ablated (no less than 5 samples in each laminae). Furthermore, four 20-mm-long line-scan measurements vertically crossed laminae, were operated in two stromatolite slabs.

3.3 Carbon analysis

Organic and carbonate carbon contents were determined using LECO RC612 multiphase-carbon analyser. Organic carbon (C_org) was measured after decarbonization with 2 N HCl. Carbonate carbon (C_carb) was calculated as the difference of directly determined total carbon (C_tot) and organic carbon (C_org).

3.4 Biomarker analysis

Biomarkers were extracted using one freshly-cut stromatolite rock sample, and the surface layer was separately removed to avoid potential diagenetic alteration and contamination. Powders were extracted stepwise with distilled dichloromethane/methanol (3:1), dichloromethane (DCM), and n-hexane, using ultrasonication. After drying and derivatizing the extracted calcite powder was dissolved. The residue samples were centrifuged and neutralized by flushing with distilled water. Further biomarker extraction was also performed by ultrasonication with DCM/methanol (3:1), DCM and n-hexane. The total organic extract, lipid fraction and combined organic extract were analysed with Thermo Trace 1310 GC coupled to Thermo TSQ Quantum Ultra triple quadrupole MS.

4 Results

4.1 Classification, texture and fabrics of Wudaoliang stromatolites

Microscopic observations revealed that two different types of stromatolites developed in the investigated outcrop: the fenestral stromatolites and the agglutinated stromatolites.

4.1.1 Fenestral stromatolite

Fenestral stromatolite is characterized by spar-cemented voids (Fig. 2a and b). Following Demicco and Hardie (1994), fenestral fabrics are referred to millimetre-scale, open or spar-cemented voids formed during syn-depositional and early diagenetic phase. Two types of fenestrae, angular fenestrae and irregular fenestrae, can be recognized. Angular fenestrae (Fig. 3a and b) are essentially elongated structures enclosed by ~ 0.5–1.0 mm long micritic sublinear structures with sharp edges. Irregular fenestrae occur as millimeter-scale, open or spar-cemented cavities without defined shape (Figs. 2b, d, 3a-b).

Fig. 2

a Vertically-cut slab of a representative fenestral stromatolite hand specimen (WDL-13). Large open voids result from late meteoric dissolution; b Vertically-cut slab of another fenestral stromatolite hand specimen (WDL-12); c Slab of agglutinated stromatolite with substantial trapped particles. Diameter of the coin is 2.5 cm; d Zoomed area of the fenestral stromatolite in Fig. 2b. Note the millimeter-scale, elongated voids vertically arrayed to the laminae; e Zoomed area (2.4× magnification) of the agglutinated stromatolite in Fig. 2c, showing trapped/bound particles in different size; f Field image of stromatolites of the Wudaoliang Group

Fig. 3

Microscopic images of Wudaoliang stromatolites. a Porous lamina (left side of the dashed line) and clotted lamina (right side of the dashed line) of fenestral stromatolite; b Elongated fenestrae (arrows) in a spongy layer, with few peloids (lines) and detrital components; c Dense peloidal lamina (the upper) and porous peloidal lamina, and the arrow showed an elongated pellet resembled to Artemia pellet by similar size (380 μm long); d Microscopic image of the thin section of Fig. 2e, showing spongy layers (with voids and spar cements) alternated with thin microcrystalline layers (no voids or spar cements), and abundant fenestral fabrics in spongy layers; e Zoomed area of rectangle zone in the Fig. 3d. Note that the angular, microspar-filled crystals (or voids) are possibly traces of the former evaporite minerals; f Zoomed area of rectangle zone in the Fig. 3d; g Pellets-rich laminae of agglutinated stromatolites, either densely packed or loosely packed; h Zoomed area of rectangle zone in the Fig. 3g. Note the almost constant size of the pellets (refer to the Fig. 3i); i Different sized (up to 500 μm) faecal pellets due to cross-section (cutting) effect in an agglutinated stromatolite

Laminae types of fenestral stromatolites were subdivided by the type and abundance of fenestrae into three groups (Table 1): 1) spongy layers formed by upright oriented microcrystalline sublinear micrite structures enclosing elongated fenestrae (Fig. 3a and b); 2) clotted microcrystalline laminae containing irregular fenestrae; 3) dense microcrystalline (aphanitic) layers.

Table 1 Classification of Wudaoliang stromatolites and laminae within

4.1.2 Agglutinated stromatolite

The investigated agglutinated stromatolite showed only minor fenestral fabrics (Fig. 2c). Instead, significant amounts of trapped and bound particles were observed (Figs. 2e, 3d-f). The particles could be divided into peloids, faecal pellets and framework micrites.

Peloids were microcrystalline grains without specific shape. Spherical, slightly elongated or even subangular peloids, sized from less than 10 μm to 100 μm. Peloids were generally surrounded by microspar cements, or irregularly clotted to other peloidal micrites.

Faecal pellets were composed of dark microcrystalline calcite and range in size from 80 μm to 120 μm (Fig. 3c, g, h). The length of some elongated pellets could attain 500 μm (Fig. 3i). The pellets showed no internal structures, and round to sub-round or elliptical shapes, occasionally broken into segments. Pellets were either bound and floated with the presence of other particles, or exclusively packed and stacked to form dense peloidal layers.

Framework micrite comprised micrite aggregates (sometimes including peloids and pellets) closely clotted to each other, causing the original periphery of particles to be no longer recognizable. Micrite was aggregated to form different morphological connections, which could be microscopically clotted, dendroid or thrombolitic-like aggregates.

Occurrences of irregular branching endolithic filaments, which were about 10 μm in diameter and locally penetrated the stromatolites on their surface layers, were confirmed as contaminations instead of primary fossils.

4.2 Stable carbon and oxygen isotope

Stable carbon and oxygen isotope results of Wudaoliang stromatolites are listed in Table 2 and plotted on Fig. 4. The δ¹³C values of the analysed fenestral stromatolite ranged from 2.340‰ to 4.265‰, with a mean value of 3.030‰. The δ¹⁸O values ranged from − 9.070‰ to − 6.095‰, with a mean value of − 7.394‰. Both carbon and oxygen isotopes showed slight oscillations in continuous alternating porous and dense laminae, with a mean δ¹³C value of 0.289‰ and a mean δ¹⁸O value of 0.514‰ in two adjacent laminae. However, there was no significant difference between porous (spongy) and dense (clotted) laminae. A positive covariation was evident between δ¹³C and δ¹⁸O ratios (r = 0.720, n = 36, p < 0.05).

Table 2 Stable carbon and oxygen isotope results of Wudaoliang stromatolites

Fig. 4

Cross-plot for the stable carbon and oxygen isotope results of Wudaoliang stromatolites. Data of the Miocene Wudaoliang drill core carbonates are quoted from Wu et al. (2009); data of the Miocene-Pliocene Dingqinghu and the Eocene-Oligocene Niubao lacustrine carbonates are from Rowley and Currie (2006); data of the Eocene Fenghuoshan lacustrine carbonates are from Cyr et al. (2005); and, the data on the recent travertine in central Tibetan Plateau are from Niu et al. (2013). The greyish arrow indicates a trend of increasing evaporation from the Eocene to the Miocene in the northern Tibetan Plateau

4.3 Rare earth element (REE) pattern

The ablation spot measurements of laser ablation inductive coupled plasma-mass spectrometer were heterogeneous, possibly due to the low REE concentrations in carbonate minerals. To deduce this effect, we calculated the mean REE value of ablation spots in each lamina (Fig. 5). The very similar shale normalized REE patterns of laminae indicated that the heterogeneous signals could be eliminated by overlapping signals of the individual points. Similar integrations were made on the line-scan mode data.

Fig. 5

PAAS-normalized rare earth element patterns of Wudaoliang stromatolites and reference samples. a PAAS-normalized REE patterns of samples WDL-13-A (ablation spot mode), WDL-13-D-1, WDL-13-D-2, WDL-10-C-1 and WDL-10-C-2 (line scan mode); b PAAS-normalized REE patterns of samples WDL-Carbonate, WDL-Shale, and WDL-Stromatolite. WDL-Carbonate and WDL-Shale were referred to the average REE composition of carbonate and shale of the Wudaoliang Group (Yi et al. 2008), WDL-Stromatolite was referred to the average REE composition of stromatolites in this study; c PAAS-normalized REE patterns of individually analysed laminae in sample WDL-13-A; d PAAS-normalized REE patterns of Wudaoliang stromatolites and reference samples. Banza aragonite (sample TGR21-I 1) was from Lake Tanganyika, East Africa (Barrat et al. 2000); Erbisberg travertine (Erb 1) and Wallerstein spring mound carbonate (Wa 98/8) were from Nördlinger Ries, western Bavaria, Germany (Arp et al. 2013); GRF stromatolite was from the Eocene Green River Formation, Wyoming, USA (Bolhar and Van Kranendonk 2007)

The results of stromatolite samples analysed by LA-ICP-MS were shown in Tables 3, 4, 5 and 6. The ΣREE of the stromatolites ranged from 3392 ppb to 5746 ppb (Table 3). Nine dense laminae of the sample WDL-13-A had a mean ΣREE concentration of 6286 ppb, whereas the eight porous laminae in this sample yielded a mean ΣREE concentration of 4479 ppb (Table 5). REE data of Wudaoliang stromatolites and reference samples in Table 3 were also presented in Fig. 5, normalized by the Post-Archean Australian Shale (PAAS).

Table 3 Rare earth element (REE) concentrations (ppb) of Wudaoliang stromatolites and reference samples

Table 4 PAAS-normalized REE values of Wudaoliang stromatolites and their reference samples

Table 5 Integration of rare earth element (REE) concentrations in stromatolite laminae of the sample WDL-13-A

Table 6 Integration of PAAS-normalized REE values in stromatolite laminae of the sample WDL-13-A

Except for WDL-13-D-1, the stromatolite samples showed positive Eu anomalies (Eu/Eu*_SN) from 1.006 to 1.260 (Table 4). And, 14 of the 17 laminae in sample WDL-13-A showed positive Eu anomalies with a highest value of 1.608 (Table 6). In samples WDL-10-C-1, WDL-10-C-2 and WDL-13-A, La/Lu_SN ratios ranged from 0.568 to 0.794; and, La/Lu_SN ratios ranged from 0.944 to 0.955 in samples WDL-13-D-1 and WDL-13-D-2 (Table 4). Sample WDL-13-D-1 had a positive Ce anomaly of 1.300 (Table 4). However, 2 laminae in sample WDL-13-A showed negative Ce anomalies (0.360 and 0.627; Table 6).

4.4 Carbon content and biomarker analysis

The six analysed stromatolite samples had a carbonate content ranging from 94.9 wt% to 97.0 wt%, but the organic carbon content hardly reached 0.1 wt%.

The biomarker content of the analysed stromatolites was very low and mainly as hydrocarbon biomarkers, namely normal alkanes (n-alkanes) (Fig. 6). Branched or cyclic hydrocarbons were not observed. The distribution of hydrocarbon biomarkers ranged from C₁₉ to C₃₃ n-alkanes (C₁₉ to C₃₃) with a bimodal pattern at C₂₂ and C₂₉ respectively (Fig. 6). The relative abundance of the odd-n-alkanes was predominantly higher than the even-n-alkanes in the range of C₂₅-C₃₃ (Fig. 6). Pristane and phytane appeared just in trace amount.

Fig. 6

Bimodal distribution pattern of the relative abundance of normal alkane detected in the stromatolite samples ranging from the mid-chain n-alkanes (C₁₉ to C₂₄) to the long-chain n-alkanes (C₂₅ to C₃₅). Note the predominance of odd-numbered n-alkanes over even-numbered n-alkanes in their relative abundance range in the range of C₂₅-C₃₃. Long-chain odd-numbered n-alkanes (C₂₅, C₂₇, C₂₉, C₃₁) could indicate the presence of higher plants in the vicinity of the depositional environment. See further explanation from the calculated index results in Table 7

This study calculated several indices based on the n-alkane distribution: CPI (carbon preference index; Marzi et al. 1993); ACL (average chain length; Collister et al. 1994); Norm31 (proportional abundance of C₂₉ and C₃₁ n-alkanes); Norm33 (proportional abundance of C₂₉ and C₃₃ n-alkanes; Carr et al. 2014); and the proxy ratio Paq (relative proportion of mid-chain to long-chain n-alkanes, Ficken et al. 2000). These indices were shown in Table 7.

Table 7 Calculated index results based on biomarker parameters of the analyzed stromatolites

5 Discussion

5.1 Palaeoenvironmental conditions

There were no indications of morphological microfossils such as cyanobacterial filament traces or calcified sheaths in the study area, which are common in freshwater settings (Freytet and Verrecchia 1998, 1999; Arp et al. 2001). Likewise, well calcified cyanobacterial filaments or other algal fossils were not found in previous studies (Yi et al. 2008; Zeng et al. 2011). Furthermore, no shelly fossils of eukaryotes (i.e., ostracods, gastropods, etc.) were observed in Wudaoliang stromatolites or other stromatolites of the Wudaoliang Group. However, in the argillaceous limestone and marlstone underlying and overlying the equivalent beds of Wudaoliang stromatolites, abundant ostracod (mostly Eucypris, few Cyprinotus and Youshashanian) and gastropod fossils (Radix sp. and Gyraulus sp.) were reported (Yi et al. 2000, 2008; Wu et al. 2008). Eucypris are a kind of common taxa in freshwater and brackish lake in northern Tibetan Plateau (Mischke et al. 2006; Wu et al. 2008); Cyprinotus is a taxon usually found from a saline lake (De Deckker 1981); and, Youshashanian usually appears in brackish to saline lake settings (Wu et al. 2008; Yang et al. 2006). Radix sp. and Gyraulus sp. are typical freshwater gastropods (Fontes et al. 1985; Taft et al. 2012). The total absence of all these fossils in the investigated stromatolites argues against a freshwater setting, and points to inhospitable ecological conditions during the stromatolite formation, such as high or strongly fluctuating salinity, high temperature, or toxic substances. Indeed, the only evidence of metazoa in the investigated stromatolites comes from elongated faecal pellets, which are similar to that of the brine shrimp Artemia pellets by similar size and shape (Fig. 3i; Eardley 1938; Kelts and Shahrabi 1986).

The presence of fenestral fabrics commonly reflects gas production and degradation in photosynthetic microbial mats, possibly indicating shallow sublittoral to littoral settings (Monty 1976; Hardie and Ginsburg 1977).

However, in the stromatolites investigated in this study, many of the fenestrae show angular outlines reminiscent of former evaporate crystals, while enclosed in vertically structured sublinear micrite resembling filamentous microbial mats (Fig. 3d-f). Besides, evaporate precipitation, specifically of gypsum, has been reported from a number of other stromatolites, reflecting high-salinity environments (Hudson 1970; Cater 1987; Arp et al. 2008; Allwood et al. 2013). Intercalated gypsum beds are also found between argillaceous marlstones of the Wudaoliang Group, indicating strong arid condition and intensive lake level fluctuations of the Miocene lake (Yi et al. 2000). In association with irregular fenestral fabrics, the possible calcite pseudomorphs after gypsum might present a frequently exposed supralittoral zone with high salinity. Although clear swallow-tail twin structures were not observed, the association of brine shrimp faecal pellets and co-varying stable carbon and oxygen isotopes (see below) are consistent with the interpretation of these angular traces as pseudomorphs after former evaporites.

5.2 Hydrology and evaporation

The covariation of δ¹³C and δ¹⁸O may be due to (1) hydrologically closed conditions of the lake basin and (2) mixed sampling between primary microcrystalline precipitates and diagenetic microspar, or both effects. However, there was no isotopic compositional difference between microspar-rich porous laminae and dense laminae. Thus, our data represented either complete diagenetic homogenization or primary signal during stromatolites formation. Zeng et al. (2011) measured the micritic calcites from laminae of stromatolites in the Wudaoliang Group and obtained oxygen isotope data ranging from − 9.91‰ to − 5.28‰. Our isotopic results of calcite from different laminae were in accordance with Zeng et al. (2011) and thus showed pristine signals reflecting the hydrochemistry condition during the carbonate precipitation.

The δ¹⁸O of Wudaoliang stromatolites in this study are similar to the carbonates in the Miocene Wudaoliang Group measured by Wu et al. (2009), despite a few argillaceous marlstone samples with higher δ¹⁸O values (close to 0). During the Miocene Epoch, a vast lake system prevailed in Tibetan Plateau, marked by channels connected several lake sub-basins covering Hoh Xil and Lunpola areas (Wu et al. 2008; Polissar et al. 2009). Most of the sediments in the Wudaoliang Group are remarkable by a weak covariation of isotopic composition between δ¹⁸O and δ¹³C (Fig. 4), therefore representing a well-developed through-flowing, open lake episode during the Miocene vast lake system (Wu et al. 2008). Carbonates of the Eocene Fenghuoshan Formation in the Hoh Xil Basin showed a less variable oxygen isotope composition (− 11.7‰ to − 10.3‰) and retained a mean value around − 11‰, indicating a hydrologically open lake setting (Cyr et al. 2005). An Eocene-Pliocene evaporation trend in the Lunpola Basin was discussed (Rowley and Currie 2006; Polissar et al. 2009), meanwhile, a similar process took place in the Hoh Xil Basin, which confirmed that this climate changing trend was pronounced in the Tibetan Plateau.

The presence of Wudaoliang stromatolites indicates an aridification of the Miocene lake environment, representing a lake-level low-stand period. Evaporation preferentially removed ¹⁶O from the water body, resulting in an ¹⁸O enrichment in the residual water body. The lake level fall caused by evaporation finally resulted in hypersaline conditions and temporary subaerial exposure, in accordance with the presence of fenestral fabrics, absence of eukaryotic fossils and possible gypsum pseudomorphs. Thus, for the formation of the investigated Wudaoliang stromatolites, a hypersaline lacustrine environment due to evaporation is more convincing than a freshwater setting.

Wudaoliang stromatolites show δ¹³C values ranging from 2.43‰ to 4.12‰, about 1.20‰ higher than skeletal limestones (Yi et al. 2008) and marlstones (Wu et al. 2009) of the same formation. High δ¹³C values might point to a significant photosynthetic effect in the water body. In photosynthetic process ¹²C is preferentially assimilated by primary producers, resulting ¹³C_DIC enrichment in lake water and high δ¹³C in the carbonates (Kerby and Raven 1985; Hollander and McKenzie 1991; Gu et al. 1996). Therefore, despite high salinity, a significant primary carbon production is evident for this depositional period of the lake in the Hoh Xil Basin.

5.3 Possible hydrothermal impact on stromatolites

The ΣREE values of Wudaoliang stromatolites were much lower than the skeletal carbonates from the Wudaoliang Group (Fig. 5b, Yi et al. 2008). Samples WDL-10-C-1, WDL-10-C-2 and WDL-13-A showed slightly enrichment of HREEs. But no significant enrichment of LREEs or HREEs was shown for Wudaoliang stromatolites, as low La/Lu_SN ratios were close to 1, arguing against highly-alkaline, soda lake conditions (e.g., Mono lake in USA, Johannesson and Lyons 1994). Further, higher ΣREE were detected in the dense laminae than in the porous laminae, as is the case with the sample WDL-13-A (Table 3).

Although low in Eu concentration, the Eu anomaly of Wudaoliang stromatolites was significant, as shown by its consistent presence in most of the analysed laminae (Fig. 5c). Sample WDL-13-A showed the most prominent positive Eu anomaly. Indeed, the Eu anomalies did not correlate with aluminium (r = 0.11, p = 0.664) or silica (r = 0.06, p = 0.819) contents. These Eu anomalies were thus not affected by terrestrial inputs or decomposition of Eu-enriched minerals (e.g., plagioclase, Sverjensky 1984; Bau 1991). Eu enrichment has been used to identify hydrothermal activities, because its mobility and redox state are closely related to the high temperature (Michard 1989; Mills and Elderfield 1995; Barrat et al. 2000).

For a better comparison, we plotted REE patterns for the aragonites precipitated from mixed sub-lacustrine hydrothermal fluid and lake water (Banza aragonite, Barrat et al. 2000), the carbonates affected by deep hydrothermal or hot spring activities (Erbisberg hot spring travertine and Wallerstein spring mound carbonate, Arp et al. 2013), and the stromatolites deposited in a freshwater lake (Green River Formation, Bolhar and Van Kranendonk 2007) on Fig. 5d. The REE patterns of Wudaoliang stromatolites are similar to the patterns of the reference samples as Banza aragonite, Erbisberg travertine and Wallerstein carbonate. In addition, stromatolites at other locations of the Wudaoliang Group formed only centimeter- to decimeter-thick beds, while the localized extensive thick-bedded occurrence of stromatolites at the slope section north of Wudaoliang Town, i.e. the Wudaoliang stromatolites in this study, possibly suggested a localized hydrothermal spring influx at this area. The hydrothermal origin of the fluids is evident from the positive Eu anomaly of the stromatolites. Nonetheless, a prominent mound structure was not evident in the sample site, so that the past hydrothermal fluids might have been introduced by the diffused sub-lacustrine spring or the surface hot spring runoff.

Furthermore, we compared the oxygen isotope of Wudaoliang stromatolites with the Holocene travertines in the Qiangtang Basin (Niu et al. 2013). The source water of the hot spring travertines is of meteoric origin (Tian et al. 2001; Zhang et al. 2002). A significantly elevated δ¹³C (up to 11.7‰) is found in these travertines, possibly related to a deep source of CO₂ (Tian et al. 2001; Zhang et al. 2002). Similar δ¹⁸O values of stromatolites to those of the Holocene travertines thus may also indicate a mixed origin of hydrothermal fluids with the Miocene period local rainfall, presuming the monsoon precipitation was similar to present conditions. However, uncertainty remains with this hypothesis, since the monsoon precipitation on northern Tibetan Plateau could have had different isotopic composition during the Miocene (see Polissar et al. 2009). Further studies are required to address this problem.

5.4 Surrounding ecology

C_org content in Wudaoliang stromatolites was less than 0.1 wt%; and the poor biomarker preservation was also consistent with the C_org content: there was no indication of specific microbial groups involving in the stromatolite formation.

Long-chain odd-numbered n-alkanes (C₂₅, C₂₇, C₂₉, C₃₁) indicate the presence of higher plants in the vicinity of the stromatolite-forming lake. In general, n-alkanes produced by forests may have a prominent n-C₂₉ > n-C₃₁ pattern, whereas grassland background may show a n-C₃₁ > n-C₂₉ pattern (Kuhn et al. 2010; Rao et al. 2011). ACL values around 29.53 can be characteristic for different kinds of trees (Hoffmann et al. 2013). The Paq ratio of 0.34 can indicate for a mixture of shallow surface sediments with some green algae from the Qinghai Lake, northern Tibetan Plateau (Liu et al. 2015). A Norm31 value of 0.49 is similar to values calculated from leaf samples of different species of Compositae (Guo et al. 2016) in differing altitudes, and also from montane plants and soils (Carr et al. 2014). However, these patterns may vary with respect to local vegetation, altitude and humid or dry climate (Rao et al. 2011; Zhang et al. 2017), therefore, revealing more detailed palaeoenvironment information is not clearly possible.

6 Conclusions

In summary, this study interpreted a hypersaline setting for Wudaoliang stromatolites. Spar-cemented, angular fenestrae enclosed in a vertically structured micrite framework were explained as evaporite pseudomorphs within calcifying microbial mats. The association of brine-shrimp faecal pellets, absence of other eukaryotic fossils, and covariation of stable isotopes further supported the interpretation of a frequently exposed and evaporated, hypersaline littoral setting.

The oxygen isotope data of stromatolite calcites were in accordance with previously published results, and the coordinated variation of δ¹³C and δ¹⁸O indicates a hydrologically closed lake setting. A higher δ¹³C in comparison to the Miocene skeletal carbonates suggested a significant photosynthetic activity in the lake during the growth of Wudaoliang stromatolites.

The possibility of hydrothermal inflow to the Miocene lake was supported by a positive Eu anomaly in the stromatolites. A hydrothermal inflow could also explain the localized occurrence of the up-to-15-m-thick stromatolitic deposits. There is no indication of a freshwater setting at the time of stromatolite formation.