Contrasting endolithic habitats for cyanobacteria in spring calcites of the European Alps

Abstract

It is often difficult to decide which cyanobacteria found in endolithic habitats of calcite spring-tufa deposits are present as ephemeral components of the biota or are persistent, structural elements. To answer this question, we repeatedly studied two microhabitats of contrasting calcareous tufa springs in the European Alps. Pigment extracts, fluorescence probe measurements of in situ samples and traditional microscopy confirmed the dominance of cyanobacteria over eukaryotic algae and their viability in both microhabitats. Spring Site 1 (Laas, Northern Italy) is characterized by a highly variable, moist to dry and sun-exposed waterfall tufa consisting of fibrous calcite. A segment of these deposits in the lateral flank of a grotto contained dark endolithic layers in dim light, 1–2 mm below the surface, where aggregated cyanobacterial cells were dominant but not directly attached to calcites, a potential sign of gentle endolithic dissolution rather than calcite precipitation induced by cyanobacteria. Site 2 (Mühlau, Austria), in contrast, is a moss-tufa microhabitat associated with a seepage spring situated in a shady gorge, where the targeted stromatolites consisted of bark-like sheets of friable, orange to light-brown when wet (drying violet) ‘styrofoam’- like aggregates of minute crystallites on the day-light exposed surfaces. These calcites were observed to nucleate directly on external sheaths of viable cyanobacteria trichomes. A polyphasic approach including LM, SEM, TEM exhibited a number of identical but also some divergent cyanobacteria of which two key taxa were specific for each of the two microhabitats (Nostoc and Pseudoscytonema at Sites 1 and 2 respectively). Both cyanobacterial communities characterised, by the cloning of 16S rDNA showed a dominance of mostly unknown and partly divergent filamentous cyanobacteria assigned to the order of Synechococcales. Our microhabitat study of alpine crenal calcites highlights the rather divergent biotic responses of cyanobacteria within spring tufa deposits.

Introduction

It is well-documented that cyanobacteria have played a crucial role in the oxygenation and CO₂ regulation of the Earth’s atmosphere, as well as in geochemical cycling of calcium and in the microbial precipitation of calcium carbonate since at least the early Proterozoic (e.g., Arp et al. 1999, Schneider & Le-Campion-Alsumard 1999, Arp et al. 2001, Andrews & Brazier 2005, Visscher & Stolz 2005, Kazmierczak & Kempe 2006, Altermann et al. 2006, Riding 2006, 2011, 2011a, Aloisi 2008, Couradeau et al. 2012). For cyanobacteria it is still debated whether EPS (extracellular polymeric substances) facilitate or hinder calcite precipitation, especially in microbial mats (Stal 2012). In contrast, culture-based calcification experiments with cyanobacteria under nature near conditions have shown that hetero-nucleation e.g. by bacteria associated with EPS glycoconjugates (bioligands) play an essential role for calcite precipitation, a process which was found to be most efficient within a certain range of moderate calcite saturation indices (Zippel & Neu 2011, Spitzer et al. 2015). Fan-like stromatolites of calcites deposited over longer time periods within mucilage-rich, macroscopic clonal growing cyanobacteria coenobia are well-known from spring streams and rivulets of the Eastern Alps as freshwater micro-reefs dominated by several Rivularia species (see e.g., Geitler 1932, Pentecost 1987, Obenlüneschloß & Schneider 2001, Pentecost & Whitton 2000, Rott et al. 2000, Cantonati et al. 2007, Cantonati et al. 2015).

Other cases of cyanobacteria calcification/mineral deposition in springs and spring streams have deserved much less attention (Cantonati et al. 2015, Pentecost 2017, Trobej et al. 2017). Spring-associated limestones (SAL), such as calcareous tufa, formed from waters supersaturated with respect to calcite along waterfalls and seepages, are implicitly considered as abiogenic constructive elements only, regardless of the role of spring biota (chiefly cyanobacteria, eukaryotic microalgae and bryophytes) inducing CaCO₃ precipitation (e.g. Pentecost 1991, Golubic et al. 1993, 2008, Sanders & Rott 2009, Sanders et al. 2011, Arp & Reimer 2014, Tran et al. 2019). We therefore selected two active SAL sites of the Eastern Alps divergent in respect to climate, sun exposure and periodicity to study endolithic cyanobacterial habitats out of a larger variety of micro-settings within the two sites.

Environmental settings

Spring Site 1 (Laas) is situated in Val Venosta, Province of Bozen / Bolzano in Northern Italy at ‘Sonnenberg’, a steep south-facing slope (Table 1) of the Etsch Valley. It forms a large spring cascade which emerges within a belt of tectonically sheared orthogneiss that contains up to 20% iron-bearing carbonate minerals (prevalently ankerite, ferroan dolomite) (cf., Froitzheim et al. 1997, Unterwurzacher 2001). In addition, the gneiss hosts veins mineralized by quartz, ankerite, and iron/copper sulfide ores. The site is among the driest locations of the Alps (annual precipitation of 440–530 mm) due to rain shadow from the NW (Fliri 1975). The water emerges from an open, subvertical joint located about six meters within a halfcave-like crevice. The focus is on a cliff coated by waterfall tufa laterally adjacent to a tufa overhang (Figs. 1, 2) approximately 10 m below the spring emergence. The external waterfall segment was ephemerally wett, much more seldom than the overhang, where water percolated more frequently. At the study site, the waterfall cliff showed a surface crust of grey to dark-grey springstone that becomes blackish when wet. The springstone showed laminations oriented subparallel to the waterfall surface which consists of dense, low-porous laminae of fibrous calcite to magnesian calcite, intercalated with laminae of tufa of higher porosity. Even when the studied section of the waterfall was dry at the surface, chipping off the superficial calcite lamina (~3–4 mm in thickness) revealed that underneath, the tufa together with its endoliths was moist to wet.

Table 1. Overview of study sites (for details of mineralogy see Table 4).

Variable	Site 1 (Laas)	Site 2 (Mühlau)
Longitude	10°41’04” E	11°24’56” E
Latitude	46°37’50” N	47°17’29” N
Altitude (m a.s.l)	1120	750
Exposition	S	E
Lithology of spring catchment	Phyllonitic gneiss	Alluvial-fan breccia
Structural unit of Alps	Eastern Alps (basement)	Eastern Alps (cover units)
Spring limestone deposystem	Waterfall tufa	Waterfall tufa
Hydrology/ Runoff	Strongly changing, cascade 0.1–5 L s–1	Permanent percolation, 0.1–0.3 L s–1
Annual precipitation	400–500 mm.yr–1	700–900 mm.yr–1
Macroscopic aspect of microhabitat	Laminated springstone (cement stromatolite) with endolithic black layers	Laminated springstone (cement stromatolite) coated by ‘leather-like’ biofilm
Mineralogy of spring limestone	Calcite to magnesian calcite	Calcite
Relation of cyanobacteria to springstone	Cell clusters in pores within springstone; clusters tightly embedded by calcite (micro)spar	Calcite crystals nucleating on ‘leather-like’ meshwork of cyanobacteria trichomes
Cyanobacteria area/ endolithic microhabitat	Confined to a few m2 total area; exposed to SW and interspersed with tintenstriche	A few dm2 on an overhanging tufa surface

Figs. 1–2.

Site 1 (Laas). 14 October 2014 after rainfall. (1) Overview showing subsampling locations 1–4 (black numbers); (2) close up of subsampling station 4 after sampling; white arrows indicate position of endolithic layers.

Spring Site 2 (Mühlau) is situated at the vertical right flank of a deeply incised bedrock gorge of the stream “Mühlauer Bach” in the NE of downtown Innsbruck. The groundwater supplying the spring acquires its chemistry by percolation through a thick succession of lithified, Quaternary alluvial fan deposits composed mainly of clasts of Triassic carbonate rocks (Sanders 2010). The spring is a seepage that is, however, active over the entire year. Facing toward the East and positioned approximately 2 m above the bottom of the gorge, the spring location is permanently shady and cool. The investigated stromatolites were located along the outer, permanently wet brink of an overhang, laterally adjacent to a “plug” ~1 m in height of moss tufa (Figs. 3, 4). The stromatolites comprise flexible, leather-like crusts up to 1.5 cm in thickness that slowly peel off (perhaps under their own weight, and due to occasional freezing in winter) their substrate eventually exposing bare, blue-green patches underneath (Fig. 4).

Figs. 3–4.

Site 2 (Laas). 2 December 2006 (3) Overview of spring area with moss tufa nose (left) and soft stromatolite layers, sampling area (square) and (4) aspect after sampling.

Methods

Field sampling, field measurements, pigments

At Site 1 microhabitat rock sampling and a detailed overview of the large spring cascade was undertaken in April 2008 and October 2008, with resampling of the cave area in autumn 2010, 2014 and 2015, and in spring 2016 (pigments). Black to dark blue endolithic layers were found every time in the central, episodically dry part of the waterfall tufa on the SW-facing flank of the grotto earlier excavated from the tufa deposit. Since these endolithic layers were found to be persistent in rock samples of a certain area (Fig. 1, black numbers 3, 4), a more detailed investigation of the deposits and their biological components of the rock was initiated by scraping with sterile razor blades to obtain small subsamples for identification, pigment analysis and cultivation. Breaking subsamples for determination of pigments per unit surface was necessary.

At Site 2 extensive sampling of the algal habitat and the surrounding deposits in the lower parts and the neighbouring areas along the rock wall and across a 30 cm overhanging moss tufa column was undertaken. First sampling was conducted in November 2006 (Fig. 3) from which the microhabitat investigated (Fig. 4) was selected for more intense study. This microhabitat was sampled again in 2010, once in all four seasons in 2014, in autumn 2015 and in spring 2016. Water samples were taken on three occasions, each in a different season.

Both springs were running on most occasions: whereas at Site 1 the microhabitat with the endolithic communities was found superficially dry once, the microhabitat at site 2 was never found dry. At Site 2 water had to be collected with a glass funnel fitted to a glass bottle to catch the drippling flow as close to the studied microhabitats as possible (maximum 50–100 cm apart).

Electrolytical conductivity was measured in situ with a LF 91 conductimeter (WTW, Weilheim, Germany), and pH with a Seven-Go pH-meter equipped with an Inlab-Pure-Pro probe (pH and temperature, Mettler-Toledo, Greifensee, Switzerland). Dissolved CO₂ (causing pH reduction at spring mouth) was analysed from back-titration to pH 8.2 (‘minus p-value’) with 0.05 N NaOH followed by a titration to pH 4.3 (‘m-value’) with 0.1 N HCl in the field (Site 1) or immediately after return to the laboratory (Site 2). Analyses were made triplicate to check for scatter (for details see Tran et al. 2019). Major ion analyses including SO₄ ^2– followed standard methods of ion exchange chromatography (Dionex, Vienna, Austria) using gradually diluted samples. Nitrate was analysed by ion chromatography (detection limit 5 μg l^–1).

Pigment analysis was based on two approaches: (A) direct measurements with an epifluorescence probe (BenthoTorch, bbe Moldaenke, Schwentinental, Germany; precision: 0.1 μg chlorophyll a cm^–2; 6 excitation wave-lengths; 3(4)-band spectral range registration); (B) extracts of pigments made with 95% ethanol single peak measurements and in vivo scans of extracts (after 8 hrs in the dark, Shimadzu Spectro-photometer UV-VIS 2000). For epifluorescence a specific evaluation software of the manufacturer was used allowing analysis of relative portions of cyanobacteria, chlorophyceae, diatomophyceae and total pigment readings on site and from electronic stored data (see Aberle et al. 2006). Calibrations were based on a gradually diluted Rivularia suspension, and background readings using a white and a black plate, and the bare spring stones respectively. The dilution gradients were checked on total rock scrape samples filtrated by Whatman GFC filters. The projection area of the sampled rocks was analysed by contouring the outlines and analysis of the projection area in mm² set in relation to the sampling device detection area.

X-ray powder diffractometry

To check the spectrum of minerals associated with the cyanobacteria, samples approximately 0.5 grams in weight of mineral substance were analysed by X-ray powder diffractometry (XRD). The dried material was ground to a fine-grained powder and mounted with ethanol on 22 × 22 mm glass slides. High-resolution data were produced under ambient conditions on a Bruker AXS D8 Discover powder diffractometer with Bragg–Brentano θ–2θ geometry, using a Cu-tube operated at 40kV acceleration voltage and 40 mA gun current. The device was equipped with a primary beam Ge-monochromator and an LYNXEYE™ silicon strip detector. Data acquisition was performed in the 2θ range between 2° and 70° using a step width of 0.01° and a counting time of 1 second per step. Crystalline phase identification was done with the program Eva.Ink and the PDF4 data base of ICDD.

SEM for mineralogy

Crystal shapes and their relation to microbial fabrics were investigated by Scanning Electron Microscopy (SEM). Dry gold-sputtered samples mounted on carbon tape were inspected with a JEOL model JSM-6010 LV electron microscope with an acceleration voltage of 10 kV, in low vacuum operation mode.

Polyphasic taxonomy

A M3Z stereomicroscope (Wild, Heerburg, Switzerland) with up to 40x magnification was used to screen fresh algal samples and to select typical coenobia for further studies. In most cases it was necessary to use some 0.2 N HCl to dissolve gently the calcite before making mounts. For LM we used a Sony BMX50 microscope with 40, 63 and 100x objective lenses connected to a Progress electronic camera system linked to our PC in bright field and interference contrast illuminations. Measurements were made with a standard micrometer calibrated scale from the electronic images. We used several common identification textbooks (e.g. Komárek & Anagnostidis 1999, 2005, Komarek 2013) and specific reference studies for identification (floristic work from studies of similar habitats – e.g. Miscoe et al. 2016). A few samples cultivated in phosphate-reduced Chu 10D medium (see Whitton et al. 1991) were also studied.

For CP-SEM rock samples cells were washed and gradually dehydrated in ethanol (0% 20% 30% 40% 50% 75% and 96%) for one hour and transferred into formaldehyde-dime-thyl-acetal (FDA, dimethyloxymethane) for 24 hrs followed by another addition for 2 hrs. To determine the cell positions in relation to calcite crystals critical point drying was made with liquid CO₂ and FDA. Both CP samples and untreated samples were sputter-coated with Au-Pd and examined with a PHILIPPS XL20 SEM microscope in 2011–2014 and a ZEISS EVO10 in 2019.

Transmission electron microscopy (TEM) was performed as previously described (Kurmayer et al. 2018). Algal cells from field samples and cultures on Chu10D medium were fixed with 2.5% glutaraldehyde for 1.5 h at room temperature in 50 mM cacodylate buffer (pH=6.8), followed by fixation in 1% OsO₄ in the same buffer at 4 °C for 18 h overnight. Samples were then dehydrated in increasing ethanol concentrations, embedded in modified Spurr’s resin and heat polymerized at 80 °C. Ultrathin sections were counterstained with Reynold’s lead citrate and uranyl acetate, and viewed with a Zeiss Libra 120 Transmission electron microscope (80 kV) fitted with a ProScan 2k SSCCD camera, operated by OSIS iTEM software.

DNA extraction and 16S rDNA cloning

DNA was extracted from rock samples of both spring sites (three samples from Site 1 in autumn 2014, one sample from autumn 2013 and 2014 from Site 2) using the standard phenol-chloroform extraction procedure (Franche & Damerval 1988). Using the cyanobacteria selective 16S rDNA primers published by Taton et al. (2003), PCR products of 1673 bp were obtained as described previously (Kurmayer et al. 2018). The PCR products were purified and cloned into pJet vector (Thermo Fisher Scientific) according to standard procedure. Individual PCR products were pooled and in general more than hundred clones have been selected randomly and digested using MspI according to the manufacturer’s instructions. Clones showing restriction length polymorphism (RFLP) were sequenced from forward and reverse using the pJet vector primers resulting in 32 environmental sequences for each of the two sites. Sequences were first compared using BLASTn (Zhang et al. 2000, using the nucleotide database excluding environmental sequences), and hits in BLASTn with highest sequence coverage were recorded. Sequences were submitted to DDBJ/EMBL/Genbank under the accession no. MT029292-MT029315 and MT639893-MT639932 (Suppl. Table 1).

For a more refined taxonomic identification the obtained genotypes were aligned with reference sequences obtained from the Cydrasil – database containing 980 sequences (Roush et al. 2018; https://github.com/FGPLab/cydrasil, version 15_October_2019) using Clustal Omega for multiple sequence alignment. Sequences from the Cydrasil database were controlled for sequence length and sequencing direction. Duplicate genotypes and genotypes containing too many undetermined nucleotides were removed. Together with more recently amended genotypes an alignment of 903 sequences (1060 bp) was created (Suppl. Table 2). The phylogenetic tree was generated using MEGA X (Kumar et al. 2018, version 10.0.5) with maximum likelihood and 1000 statistical bootstrap values. A General Time Reverse Model (GTR +G +I) with a discrete Gamma distribution showed the highest log likelihood (–76597.668) and was used to model evolutionary rate differences among sites (5 categories (+G, parameter = 0.613; +I, parameter = 0.071). All positions with less than 95% site coverage were eliminated, i.e., fewer than 5% alignment gaps, missing data, and ambiguous bases were allowed at any position (partial deletion option). There were 848 positions in the final dataset.

Results

Spring water features

Despite different hydrological conditions within the catchments, both springs sheded ambient-temperature waters, i.e. there was no input of thermal water (Table 2). Although air temperature may cause some warming in summer, at the investigated sites, water temperature never reached air temperature. Both springs sheded Ca-Mg-bicarbonate-sulfate waters (see Table 2). There was some pH depression by excess CO₂ at both spring mouths causing a pH well-below 8, although the spring waters were well-buffered.

Table 2. Physico-chemical parameters of study sites (Site 1: 5. April 2018; Site 2: 3. March 2016; SI calcite saturation index).

	Site 1	Site 2
Distance of study site from spring emergence (m)	10.0	6.0
T (°C)	11.4	8.7
pHin situ	7.05	7.82
Conductivity (μS/cm25°)	1860	704
Ca2+ (mM/L)	13.22	4.61
Mg2+ (mM/L)	14.07	3.37
K+ (mM/L)	0.39	0.03
Na+ (mM/L)	1.04	0.08
HCO3 – (mM/L)	6.46	3.29
CO2 (mM/L)	0.93	0.18
Cl– (mM/L)	0.52	0.04
SO4 2 – (mM/L)	19.92	4.42
Total P (PO4-P μg/L)	n.a.	1.5
NO3-N (μg/L)	1247	648
SI	+ 0.051	+ 0.180

At Site 1, the spring water is oxygen undersaturated at emergence and rich in ferrous iron that is rapidly precipitated by iron-oxidizing bacteria under microaerobic conditions first. Only a few meters downstream, where the water descends a waterfall, all of the iron was consumed, the water was oxygenated, and calcite precipitation started (cf. Sanders et al. 2010, 2011). The discharge of the spring was estimated to be about 0.5–1 L s^–1 at peak times. At both springs, discharge and the degree of wetting of the slope of the limestone-depositing part of the deposystem were highly variable with season, in particular by highly variable rainfall episodes during summer and autumn at Site 1. A large part of the water emergent at the spring mouth re-percolates into the body of spring limestone, and partly re-emerges farther downslope. Similarly, water flow over the ‘waterfall cliff’ is highly variable over seasons and years. The spring of Site 2 acquires its chemistry by percolation through Quaternary alluvial fan deposits composed mainly of clasts of Triassic carbonate rocks. At Site 2, the studied stromatolite was wet almost all year round, although elevated quantities of water percolated during extended rain periods and periods of snow cover. At both springs, nitrate values were in the medium range (see Table 2).

Pigments

The epifluorescence measurement in situ and ex situ (sample based, Table 3), along a gradient at Site 1, showed cyanobacteria dominance for 5 of the 7 subsamples including the endolithic sample 7 (one of several replica shown), since the outer areas of the transect within the shoulder of the grotto were partly dry. In one case green algae dominated (sample 1 small rivulet, Zygnema) and in another case a dry rock sample (2) dominated by diatoms was sampled. At Site 2 all samples regardless of position within the approx. 2 m² area were cyanobacteria dominated showing endolithic black layers (Table 3). On several occasions an additional check of algae and cyanobacteria viability was made using 95% ethanol extracts of which three key examples are shown in Fig. 5 (spectra from UVA (300 nm) to the far end of the visible spectrum at 800 nm). There was a distinct chlorophyll peak in all three in vivo scans in the red (665 nm), whereas only for sample 3 at Site 2 a distinct chlorophyll a peak in blue (435 nm) was detected. For Site 1 (samples 3a and 3b), in contrast, the almost identical broad band peaks between 360 and 440 nm, are likely to be attributed to an overlay of both chlorophyll (in the visible part with a peak at 430nm) and scytonemin (390 nm peak). The extracts of Site 1 detected high scytonemin levels since ethanol tends to stabilize this UV shielding pigment (see Ekebergh et al. 2015). High PAM rates found in a supplementary study by Herburger et al. (2016) are likely to be related to high chlorophyll / unit surface area quantities found here. In addition, remarkable quantities of the pigment phycoerythrin, an indicator of adaptation of the cyanobacteria to low light at the shaded Site 2, was also detected in this study.

Table 3.

Pigment content of rock samples obtained by epifluorescence probe (columns 2–4, sum 1) and ethanol extraction (area in cm² and sum 2) (units in columns 2–5 and 7 in μg chlorophyll a L^{– 1}). * endolithic sample

	CYA	CHL	DIA	Sum 1	Area	Sum 2	Comment
Site 1
1	2.8	4.6	2.3	9.3	3.1	16	green
2	0	0	3.0	3.0	2.5	4	dry
3	3.0	2	0.2	5.2	3.5	10	wet
4	4.2	1.2	0.7	6.1	4.4	9	wet
5	6.0	0	1.4	7.4	1.7	16	brown
6	2.4	0	0.8	3.2	3.0	16	brown
7*	0.3	0	0	0.3	3.5	4	black
Site 2
1	3.5	0	2.1	5.6	3.8	5	violet
2	3.0	0	0.7	3.7	3.1	5	dark brown
3	1.2	0	0.6	1.8	1.3	13	clear brown
4	2.5	0	0.8	3.3	1	23	brown
5	2.1	0	0.8	2.9	1.4	5	brown

Fig. 5.

Spectrophotometric scans (UV A and visible spectrum) of ethanol extracts from stromatolites. Blue and red lines: Site 1:epi- and endolithic layer from subsample 3 respectively; grey line: Site 2: total stromatolite.

Stromatolites

Site 1 cement stromatolites (or laminated waterfall tufas) (Fig. 6) were characterized by black to dark blue up to 1 mm thick endolithic layers. The endolithic layers were present within patches at least 2 m² in area of the central, frequently dry falling part of the waterfall tufa adjacent to the eastern overhanging tufa (dubbed ‘overhang’). In rock samples, these endolithic layers were found to persist over at least ten years. The most striking habitat character was the highly variable waterflow–ranging from dry (e.g., autumn 2008) to high discharge (e.g. 5 L s^–1 in autumn 2015). In addition, e.g. in autumn 2015, rainfall before sampling had resulted in a thin water film along the outer part of the overhang (Fig. 1). The surface of the waterfall tufa was coated by patchy and variably thick, blackish cyanobacterial biofilms (‘Tintenstriche’), but occasionally also by fan-like pseudobranched Dichothrix gypsophila within the uppermost 1 mm layer and bryophyte tufts. The targeted endolithic layers were best differentiated in the tufa remote from the bryophyte tufts. In vertical section, samples of laminated waterfall tufa, essentially cemented stromatolites, displayed one or two black endolithic layers of variable thickness.

Figs. 6–8.

Macroscopic aspects of stromatolites. (6) Site 1 (Laas), black endolithic layer; (7) Site 2 (Mühlau) wet young bark-like stromatolite, (8) Site 2 dry styrofoam-type aspect with smooth brown or slightly violet skin-like cover.

At Site 2, the active stromatolite surface displayed a partly translucent, orange to light yellow hue (Fig. 7) that turned grey when dry (Fig. 8) and was of a consistency reminiscent of soft leather or tree bark. The light yellow to orange active layers is interspersed with a few dark-brown patches dominated by Scytonema myochrous (single dark threads in Fig. 7). When dry, the crusts keep shape and show violet pigmentation sometime after desiccation (Fig. 8).

Mineralogy for Site 1 (Laas), suggests that the fibrous crystals (see petrographic thin section descriptions below) may be magnesian calcite containing more than 4 mol% of MgCO₃ in the crystal lattice (Table 4). The Mg concentration in calcite was quantitatively determined by the amount of lateral shift of the 2 Theta-value of the main XRD peak (Goldsmith et al. 1961, Paquette & Reeder 1990). Over a total of seven samples, the MgCO₃ content ranged from 2.2–4.3 mol%, i.e., the Laas spring precipitates cluster around the limit to magnesian calcite (some just within, some just below the Mg calcite field). From Site 2 (Mühlau), four samples of stromatolites were X-ray analysed. For all samples at Site 2, the rhombohedral shape of crystals indicated that these are low-magnesian calcite with a Mg content well-below 4 mol%. At both sites, except for (magnesian) calcite, no other spring-related mineral was identified.

Table 4.

X-ray powder diffraction (XRD) based determination of mineralogy of spring precipitates. See text for further explanations

Site	Sample code	Mineralogy	MgCO3 content	Remarks
Site 1 (Laas)	ÜBH 4	Low magnesian calcite	3.6 mol%	Variations in MgCO3 content may result from variations in water chemistry and/or from early diagenetic overprint
	ÜBH 5	Magnesian calcite	4.3 mol%
	ÜBH 6 (outer fringe)	Low magnesian calcite	2.2 mol%
	ÜBH 6 (bulk of several tufa laminae)	Low magnesian calcite	3.4 mol%
	ÜBH 7	Low magnesian calcite	3.0 mol%
	ÜBH 8-A	Magnesian calcite	4.2 mol%
	ÜBH 8-B	Low magnesian calcite	2.8 mol%
Site 2 (Mühlau)	MK4	Low magnesian calcite	–	Rhombohedral shape of crystals and absence of main-peak shift in XRD indicate negligible MgCO3 content
	MK8	Low magnesian calcite	–
	MK9	Low magnesian calcite	–
	MK10	Low magnesian calcite	–

In petrographic thin sections, the waterfall tufa from Site 1 consisted of subparallel, dense laminae of radial-fibrous calcite to magnesian calcite intercalated with laminae of more porous tufa and the black endolithic layers (Fig. 9). The endoliths were tightly embedded into calcite (Fig. 10). Similarly, in CP SEM, coccoid endolithic cells (disintegrated Nos-toc sp.?) were found tightly packed by calcite fabrics (Figs. 11, 12).

Figs. 9, 10.

Springstone microfacies of waterfall tufa (Site 1, Laas) showing two blackish (a thinner and a thicker) endolithic (Nostoc-related) lamination, (active surface of waterfall tufa on top; plane-polarized light).

Figs. 11, 12.

Stromatolite Site 1 (Laas) close-up CP SEM of endolithic black layer 15 October 2015, showing multiple depressed (Nostoc-related) coccoid cells and empty spaces from cell imprints surrounded by weathered calcite.

In petrographic thin section, the Site 2 stromatolites are characterized by thin calcified laminae separated by a wide interlaminar space, resulting in an extremely porous fabric (Figs. 13, 14, 15) (‘styrofoam and spidernet fabrics’, Tran et al. 2016). The biotic component of the active stromatolites formed a dense felt-like network of cyanobacterial trichomes (Fig. 16) with little calcification on top of which a granular layer is combined with large amounts of mucilage (exopolymers). The calcite crystals, commonly seen as equant to elongate rhombohedra with rounded edges (Figs. 17, 18), directly nucleated and grew on the cyanobacterial filaments (for further details see key cyanobacterium Pseu-doscytonema below).

Figs. 13–15. Microfacies of (Pseudoscytonema-related) styrofoam-type tufa from Site 2 (Mühlau), overview and gradual close ups.

Figs. 16–18. Stromatolite SEM from Site 2 (Pseudoscytonema-related), (Mühlau), (3 different close ups, CP SEM only in 16 and 18) with variable density of calcifications and size and density of crystallites.

Key cyanobacteria

Nostoc spp.: At Site 1 we found Nostoc spp. in both epi- and endolithic habitats showing a wide variation of growth form types (and live cycle stages) (Figs. 19–21). The aspects in the endolithic layers comprised mostly small sized spherical colonies with gradually fragmenting cell chains (Fig. 20), sometimes completely disintegrating into single cells (Fig. 19). The sheath was mostly orange brown, often with variable scrobiculated surfaces contrasting to the blue (rarely brown) pigmented cells. The cell shape varied from spherical to barrel shaped or slightly longer than wide (before division) within the size range 3–6 μm, arthrospores mostly larger than vegetative cells, most heterocysts circular and around 5 μm in diameter. Hormogonia cell size was 3–4 μm with sometimes the terminal cells longer than wide and lightly attenuated. There was normally a wide and laminated mucilage around trichomes (Fig. 21). It grew well on Chu 10D agar (Whitton et al. 1991) at first, but gradually disintegrated evidently by nutrient starvation (see SEM Figs. 22–25). From one of the batch cultures given to the University of Liège culture collection, a pure culture was established (Willmotte Annick, personal communication). Since Nostoc is a species rich taxon, it is possible that several morphotypes are represented here.

Figs. 19–21.

LM of Nostoc from field samples Site 1 (Laas) and a culture respectively. (19) ex situ aspect of coenobium single cell status in autumn 2008, (20) endolithic sample showing roughly trichal status with one distinct heterocyst (arrow), (21) aspect of growth on agar with some gradually disintegrating trichomes (left upper corner).

Figs 22–25.

TEM of Nostoc sp. cultivated on CHU 10 agar, isolated from Site 1 (Laas). Fig. 22 overview illustrating the arrangement of a few cells into a trichomes and cell cluster respectively; Fig. 23 detail with several connected cells; Fig. 24 two cells connected by a septum; Fig. 25 detail with internal organization of a cell with multiple parallel submarginal thylakoid membranes and central carboxysomes. Abbreviations: Cg cyanophycin granule, Cs carboxysome, Sp septum, Sh sheath, T thylakoids.

The TEM’s of Nostoc (Figs 22–25) show various aspects: an overview with a filament fragment and an irregular cluster of cells in Fig. 22, a close up in Fig. 23 of two groups of cells, where in the cells on the right individual electron dense carboxysomes and cyanophycin granules are visible; a segment of a trichome in Fig. 24 with two connected cells in the centre and large cell in good shape in the lower left corner sharing its sheath with an almost collapsed sister cell, sacrificed to maintain the larger one. The two cells sharing the septum, share both white empty areas and electron dense ones, some inside the white areas, which all are probably polyphosphate granules. In Fig. 25, the largest close-up shows the arrangement of thylakoid membranes in the cell periphery, as well as carboxysomes in the cell centre. Plenty of diffuse exopolymers covering almost all cells are likely to facilitate endolithic growth. (Some of the variability of the cell shapes and sizes (especially in Figs 22, 23) may come from the ultrathin section that is only about 60 nm thick).

Pseudoscytonema sp.: At low magnifications the active stromatolite showed a dense network of translucent and irregular bent minute trichomes forming a thin soft yellow to orange mucilaginous (first 1–2 mm thick) bark-like layer (Fig. 7). LM features: Single trichomes extended with occasionally false-branching, one firm transparent single sheath layer and somewhat variable terminal cells (inflated or snake head-like, see arrow in Fig. 26). Sometimes with several shorter barrel-shaped subapical cells (meristematic zone). Occasional vacuolised cell content – with cross walls straight or slightly constricted. Trichome diameter (without sheath) 2.1–2.8–3.3 μm (sheath envelope diameter 2.1–3.8(4) μm) cell length (2.1)–4.6–8–(10) μm –– mat forming taxon and terminal inflated structures bent outward of the mat. Single trichome undulating or irregular bent interwoven into networks.

Fig. 26–28.

Pseudoscytonema sp. LM of field aspects. (26) several uncalcified trichomes, one with inflated terminal cell (arrow); (27) filament with primary calcification aspect (upper arrow), lower arrow pinpoints to uncalcified parts with non-inflated terminal cell; (28) overview showing one rhombic calcite crystals with trichome trace.

The role of Pseudoscytonema in the peculiar calcifications at Site 2 was striking for us. We observed repeatedly that the trichomes of this taxon had layers of minute calcite crystals on their sheath (presumably early stages of calcification, Figs. 27), wheras other minute calcite crystals showed traces of single cyanobacteria trichomes (arrow in Fig. 28). The network of cyanobacterial trichomes seemed to hold the crystals together at least in an early stage of calcification (Fig. 17). Individual crystallites ranged from less than one micron to a few microns in size (Fig. 18), but they coalesced into larger aggregates (Fig. 17) sometime also visible in optical light microscopy. Calcite crystals consist of: (a) arrays of subcrystallites parallel to each other and displaying the trigonal symmetry of calcite; (b) more commonly, crystallites subsequently forming rhombohedra, typically with rounded edges (Figs. 18, 28). Both types of crystallites are clustered into aggregates up to a few tenth to hundreds of microns in size (Fig. 17).

Hormogonia had mostly square-shaped cells (3x3(4) μm). Cell division type Phormidium-like, i.e. single cell division without anticipation of another division (not Oscillatoria-like) and trichome endings with a somewhat variable subapical meristematic zone (acc. to Komarek & Anagnostidis 2005, see Fig. 26) allow a classification into Micro-coleaceae stat. nov. We see similarities to the growth form type of Plectonema capitatum described by Jaag (1935) from spray zones of the Lunzer Oberer Seebach, later reclassified into the genus Pseudoscytonema by Komárek & Anagnostidis (2005).

TEM features of Pseudoscytonema (Figs. 29–31) show segments of straight intercalary cells of trichomes in Fig. 29, parts of one subterminal somewhat inflated and one terminal cell in Fig. 30, and a cross-section of a trichome in Fig. 31. All cross sections and longitudinal sections indicate an irregular positioning of thylakoids spread over a large portion of the cell diameter, a typical feature of coccoid cyanobacteria (Mares et al. 2019). The cross walls showed minute gaps in the septa between cells–with uniform size (white circles, Fig. 29). Along the cell wall a dense layer of arrays, oriented perpendicular to the cell wall, was observed (arrows)–the function of this structure remains obscure, potentially they are related to locomotion. The cell walls are thick, 0.5–1 μm consisting of three layers, a solid thin murein layer (with some bubbles), a solid homogeneous appearing sheath layer (0.1–0.2 μm) and a diffluent (mucilaginous) wider external layer (0.5 μm) (Figs. 29, 31).

Figs. 29–31.

TEM of Pseudoscytonema sp. isolated from Site 2 (Mühlau). Fig. 29 detail of intercalary cell with thylakoid membranes spread throughout the cell body; in the cell cortex a dense layer of arrays oriented perpendicular to the cell wall were observed (arrows), pore-like connections between the individual cells are marked with circles; Fig. 30 terminal cell embedded in a massive sheath; Fig. 31 cross section through individual cell. Abbreviations: cw cell wall.

Clone library results

A total of 64 environmental 16S rDNA sequences (938–941 bp) were obtained that differed maximum 14.2% in nucleotide similarity. 31 genotypes were obtained from Site 1 (Laas, n=32 sequences) and 32 genotypes from Site 2 (Mühlau, n=32 sequences) (Suppl. Table 1). Using BLASTn for Site 1 the majority of genotypes were found similar to the thin filamentous cyanobacteria classified among the Synechococcales, such as Leptolyngbyaceae (Komárek 2016, Mai et al. 2018), i.e. Leptolyngbyaceae cyanobacterium LD4 4600 EK (n=13), Cyanobacterium NIES-36 (n=9), Oscillatoria sp. N9DM (n=3), Tildeniella sp. T16 (n=2) and Nostocales cyanobacterium LEGE 06106 (n=2) as well as Plectonema cf. radiosum LEGE 06105 (n=1) and Chroococcus sp. CCALA 702 (n=1). Similarly the majority of genotypes from Site 2 were found more similar to the same filamentous cyanobacteria classified under Synechococcales, such as Tildeniella sp. T16 (n=26) and Leptolyngbyaceae cyanobacterium (n=2) as well as Trichocoleus sp. ACSSI 301 (n=1) and Radiocystis sp. JJ30-12 (n=1). Using the BLASTn derived taxonomic assignments rarefaction curves have been calculated showing that for both sites the sampling depth approximated saturation. For both sites the genotypes of the Leptolyngbya like cyanobacteria were found dominant while coccoid and filamentous cyanobacteria from other orders (Chroococcales, Nostocales) occurred occasionally (Suppl. Fig. 1).

To achieve a more robust phylogenetic assignment all cloned genotypes were integrated into a maximum likelihood phylogenetic tree analysis using 839 taxa from the Cydrasil database. The majority (n=58 out of 64) of genotypes from both sites were grouped within the large clade of the order of Synechococcales related to thin filamentous cyanobacteria generally named “Leptolyngbya” (Fig. 32, Suppl. Fig. 2). While one clade was dominated by genotypes from Site 1 (Laas), a second clade was dominated by the genotypes from Site 2 (Mühlau). Both clades were phylogenetically most closely assigned to representatives of the family of Oculatellaceae (Mai et al. 2018). A few genotypes from Site 2 were phylogenetically grouped together with representatives of the family of Leptolyngby-aceae and Trichocoleaceae (Mai et al. 2018). Less frequently observed genotypes were included within Chroococcales (Chroococcus_sp_CCALA_702, Radiocystis sp_JJ30_12), and Nostocales (Anabaena sp. 90, Trichormus_variabilis_HINDAK_2001_4, Rivularia_sp_BECID10), (subtrees not shown).

In summary the genetic results from the two sites showed a rather homogeneous cyanobacteria community composition and were both dominated by filamentous cyanobacteria assigned to various Leptolyngbya genotypes (Table 5). Notably Site 1 (Laas) differed by the occurrence of several Nostocales genotypes. However some macroscopic taxa of cyanobacteria found linked to the endolithic layers (e.g. Dichothrix gypsophila at Site 1) have not been recorded from clone libraries, whereas others found even common in other microhabitats of both springs (i.e. Petalonema alatum Berkely ex Kirchner 1898 and Scytonema myochrous) were not considered in LM and geological assessments here by the strictly microhabitat related focus of this approach.

Table 5. Putative cyanobacteria taxa and related type of record (L light microscopy, M 16S rDNA analysis, T Transmission electron microscopy).

Taxa	Site 1(Laas)	Site 2 (Mühlau)	Microhabitat / comment
Ammatoidea sp.	L		Epilithic, trichomes with brown thick sheath, whorls of filaments
Aphanocapsa sp.	L		Freshwater
Chondrocystis dermochroa	L		Epilithic, brown envelopes
Chroococcus sp.	M	L	Freshwater
Cyanothece sp.		M	Freshwater
Dichothrix gypsophila	L		Frequent Endolithic across uppermost rock layers; thick multiple sheath layers, branched chandelier like endolithic and epilithic
Gloeocapsa alpina	L		Epi and endolithic, dark blue to black
Gloeocapsa rupestris	L		Epilithic, multiple orange to brown envelopes
Gloeocapsa violascens	L		Epi dark blue
Gloeothece cyano-chroa	L		Endolithic and epilithic light dark violet mucilage multiple or wide layers
Gloeothece membran-acea		M	Freshwater
Leptolyngbya spp.	L, M	L, M	Unresolved taxa
Nostoc sp.	L, M,T		Epi- and endolithic mucilage spheres in situ; multiple stages, whorls to typical spheres
Pseudoscytonema sp.		L, T	Dominant taxon of leather-like loose attached covers; end cells seasonal inflated on margin of coenobia; netzwork of trichomes rarely branched
Rivularia sp.	M		Frehswater, but also described from brackish/ marine water
Schizothrix tinctoria		L	Dark flesh red trichomes single or multiple
Schizothrix sp.		L	Single filaments together with Pseudoscytonema
Trichocoleus deserto-rum	M	M	Described from desert soils
Trichormus variabilis	M		Common in soils

Discussion

Our results showed again that active limestone-depositing springs offer multiple suitable micro-niches for growth and development of cyanobacteria (see Cantonati et al. 2015). The dominance of cyanobacteria (in relation to eukaryotic algae) in two stromatolite microhabitats was confirmed here by both in vivo fluorescence (Table 3) and spectrophotometric analysis of rock sample extracts (Fig. 5). From a methodical point of view, our results show that in vivo fluorescence analysis correlates well to the spectrophotometric analysis of extracted pigments, as shown by Harris and Graham (2015) especially for low and moderate chlorophyll concentrations. The moderate concentration ranges of 5–26 μg chlorophyll per cm^–2 at both study sites compare well, for instance, with strictly endolithic cyanobacteria from Antarctica dry valleys (Colesie et al. 2016). Pigment analyses demonstrated the viability of cyanobacteria in the endolithic layers. A more complex community of cyanobacteria (not only the key taxon Nostoc) at Site 1 seemed to allow growth and persistence of photosynthetic organisms under less favourable endolithic situations at this highly sun exposed, periodically hot and often desiccated site. These were presumably protected from excess UV within the endolithic cavities but kept still UV shielding pigments with an almost identical pigment spectrum as the related surface samples (Fig. 5). It seems likely that light penetration through calcites here are sufficient to allow growth, since considerable light quantities reaching through the upper mm sheets of dolomite were attested for rocks in Alpine areas of Switzerland (Horath et al. 2006). In contrast at Site 2, the specific growth form of “Pseudoscytonema” forming extensive mats of interwoven trichomes showed no UV shielding pigments, potentially not required for excess light or UV protection in such a shady setting. However, Pseudoscytonema was found to be adapted to grow at extended light and temperature ranges with a growth optimum of 16 °C (Herburger et al. 2016). The growth form and calcification enable it to colonize perpendicular rock surfaces favoured by gentle percolation of seeping waters supersaturated with calcium carbonate.

At Site 1, largely abiogenic calcite deposition must take place mainly during episodes of peak floods from heavy rain in the spring. The endolithic cyanobacteria probably started to grow at the sun-exposed surface of the tufa when it was wet but became trapped beneath a lamina of calcite growing by abiotic precipitation. Whether the identified cyanobacterial taxa, Nostoc and Gloeocapsa, however, are effective in calcium carbonate dissolution to keep on growing in the rock cavities remains unclear although, for longer persistence, this would be necessary.

Site 2 is an example where the spring biota (chiefly cyanobacteria and bryophytes) exclusively induce CaCO₃ precipitation. At this shady and perennial percolated site, surface accretion of calcite laminae was probably related to biocalcification starting with nucleation on cyanobacterial exopolymers (EPS) obviously with a clear dominance of the bioactive and shade adapted Pseudoscytonema sp. leading to highly porous stromatolite fabrics. We found all stages of calcification, from just a few crystals nucleated on algal threads to relatively dense crystal aggregates (Figs. 16–18), allowing us to dub the result as a Pseudoscytonema stromatolite. The constantly high porosity of the “styrofoam-like”, friable stromatolites beneath their surface (see Figs. 13–18) with a total thickness of 1 cm suggests that the presence of endolithic, viable Pseudoscytonema within these stromatolites was not associated with significant further calcite precipitation in the lower and more shady layers of the calcites. The faint calcification of the lower part of these stromatolites renders them prone to be episodically spalled off for instance by frost action.

The genetic diversity as revealed by cyanobacteria-selective 16S rDNA environmental cloning comprised coccoid, filamentous non-heterocystous and filamentous heterocystous cyanobacteria. Several of the described taxa are known from aerophytic sites forming biocoenoses with pronounced chemical-physical gradients varying both diurnally and seasonally. For example, the genera Cyanothece and Leptolyngbya have been described frequently from wet walls of both granite and sandstone rocks (Albertano 2012). Mats occurring in karstic springs such as observed in this study have been investigated less frequently and may be considered islands in the biogeographic sense. Indeed, for both sites a more clonal population structure of the abundant filamentous cyanobacteria generally named “Leptolyngbya” was observed. For both sites, rarefaction curves showed a saturation implying that the sampling depth was appropriate to represent the community composition. The molecular data suggest the presence of a majority of taxa distantly related to the thin filamentous cyanobacteria within the order of the Synechococcales. For part of the genotypes of both sites phylogenetic affiliation (similarity 92–93%) to Oscillatoria geminata SAG1459-8 (Jaaginema geminatum), isolated from a factory cooling tower in Germany has been found. Other more distant affiliation with desert soil alga Trichocoleus desertorum (Mühlsteinova et al. 2014) and representatives of the Leptolyngbyaceae (Mai et al. 2018) are only pinpointing genetically related taxa (Fig. 32). Even the presence of several genotypes related to Nostoc sp. have to be seen in the light of the wide occurrence of this taxon in soil crusts (e.g. Williams et al. 2016), and as endoliths e.g. in dolomite as confirmed more recently for the European Alps (e.g. Sigler et al. 2003, Horath et al. 2006).

The contrast of the two habitats given by the site-specific exposure, microclimate, different levels of buffering capacities (higher conductivity at Site 1) and especially by different runoff conditions (more episodical at Site 1, permanently percolating at Site 2, Table 1) are probably not the major reasons for the difference of stromatolite types. The environmental settings/conditions are likely to allow for abiogenic calcification in both springs since the Langelier Index is positive although not very high (Table 2). Both environments can be ranked into the environments/habitats where photosynthesis-induced cyanobacterial calcification is possible according to Arp et al. (2010) and Brinkmann et al. (2015). It is likely that both the microhabitat conditions and the adaptive capacities of the taxa involved are important to determine styles of biocalcification (see Merz-Preiß & Riding 1999, Pentecost & Whitton 2000). At Site 1, the presence of a permanently wet endolithic habitat underneath the episodically dry-falling surface of the waterfall tufa protects against drying, temperature extremes and extreme illumination, sustaining the endolithic layers. The now endolithic cyanobacteria, however, may have started to grow at the light exposed surface of the waterfall tufa when it was active, but became trapped beneath a calcite lamina, and then comprises an endolithic community that may not be very long-lived.

Fig. 32.

Expanded subtree out of Maximum Likelihood phylogenetic tree shown in Suppl. Fig. 2 with reference sequences of the order of Synechococcales including the families Leptolyngbyaceae, Oculatellaceae, as well as Trichocoleaceae. Clone sequences from Site 1 (Laas) in blue font, and from Site 2 (Mühlau) in red font. Representative genotypes used in Mai et al. (2018) are marked yellow (Leptolyngbyaceae), green (Trichocoleaceae) and pink (Oculatellaceae). For readability reasons every second genotype is indicated only. A full list of (reference) genotypes is given in Suppl. Table 2.