Introduction
Primary production is a fundamental ecosystem process that supports food webs and ecosystem energy flow. In environments where environmental stressors are common (e.g. saline conditions, intense ultraviolet (UV) radiation), microbial mats, consisting mainly of cyanobacteria but also including green algae and moss, are often foundational components of the autotrophic community. Microbial mats typically occur as laminated millimetre- to centimetre-thick colonial communities on the surface of soils and sediments in extreme environments such as deserts, hypersaline ponds and hydrothermal hot springs (Stal & Caumette Reference Stal and Caumette2011, Prieto-Barajas et al. Reference Prieto-Barajas, Valencia-Cantero and Santoyo2018). Microbial mats are self-sustaining communities that govern nutrient cycles and create a habitat supporting a diverse community of organisms (Davey & O’Toole Reference Davey and O'Toole2000, Guerrero et al. Reference Guerrero, Piqueras and Berlanga2002, Bolhuis et al. Reference Bolhuis, Cretoiu and Stal2014, Prieto-Barajas et al. Reference Prieto-Barajas, Valencia-Cantero and Santoyo2018). In the polar desert of the Antarctic McMurdo Dry Valleys, microbial mats are present within and along the margins of glacial-melt streams, lakes, wetlands, snow-influenced soils and even surface lake and glacial ice (Paerl & Priscu Reference Paerl and Priscu1998, McKnight et al. Reference McKnight, Niyogi, Alger, Bomblies, Conovitz and Tate1999, Power et al. Reference Power, Salvatore, Sokol, Stanish and Barrett2020, Stone et al. Reference Stone, Devlin, Hawes, Welch, Gooseff and Takacs-Vesbach2024). The microbial mats of this region are dominated by cyanobacteria but also contain other autotrophic and heterotrophic prokaryotic taxa (Van Horn et al. Reference Van Horn, Wolf, Colman, Jiang, Kohler and McKnight2016), as well as green algae, diatoms, mosses and invertebrates (e.g. nematodes, rotifers and tardigrades; Alger et al. Reference Alger, McKnight, Spalding, Tate, Shupe and Welch1997, Simmons et al. Reference Simmons, Wall, Adams, Ayres, Barrett and Virginia2009). The highly variable and physiologically stressful conditions of Antarctica (e.g. low nutrient availability, freeze/thaw cycles and intense UV radiation; Hawes et al. Reference Hawes, Howard-Williams and Vincent1992, Vincent et al. Reference Vincent, Downes, Castenholz and Howard-Williams1993) require key adaptations of resident biota to these harsh environments.
The terrestrial surfaces of the McMurdo Dry Valleys have been influenced by the waxing and waning of glacial ice, which left behind glacial drifts consisting of soil and rock material (Bockheim et al. Reference Bockheim, Prentice and McLeod2008). This landscape history has created significant variation in inorganic phosphorus (P) availability across the Taylor Valley in particular, where differences in till lithology and weathering rates have resulted in areas containing high and low P (Bate et al. Reference Bate, Barrett, Poage and Virginia2008, Heindel et al. Reference Heindel, Spickard and Virginia2017, Reference Heindel, Lyons, Welch, Spickard and Virginia2018). Previous studies have demonstrated a link between P availability and nitrogen (N) fixation in aquatic and terrestrial systems, where high-P environments favour N-fixing organisms (Howarth et al. Reference Howarth, Marino and Cole1988, Andersson et al. Reference Andersson, Höglander, Karlsson and Huseby2015, Tierney & Wurzburger Reference Tierney and Wurzburger2024). Moreover, we suggest the variation in P availability across Taylor Valley may influence the distribution and activity of terrestrial microbial mats, as these communities are known to contain active N-fixers (Coyne et al. Reference Coyne, Parker, Lee, Sohm, Kalmbach and Gunderson2020). Previous studies have also demonstrated that the abundance and composition of microbial mat communities are influenced by the local geomorphology and hydrology (McKnight et al. Reference McKnight, Niyogi, Alger, Bomblies, Conovitz and Tate1999, Stanish et al. Reference Stanish, Nemergut and McKnight2011, Kohler et al. Reference Kohler, Stanish, Crisp, Koch, Liptzin, Baeseman and McKnight2015, Levy et al. Reference Levy, Cary, Joy and Lee2020), but no work has systematically examined the influence of geochemical variation in nutrient availability (e.g. Barrett et al. Reference Barrett, Virginia, Lyons, Mcknight, Priscu and Doran2007, Bate et al. Reference Bate, Barrett, Poage and Virginia2008, Welch et al. Reference Welch, Lyons, Whisner, Gardner, Gooseff, McKnight and Priscu2010) on microbial mat distribution, composition and activity.
In many ecosystems, microbial mats make important material (e.g. carbon (C) and N fixation) and functional (soil cohesion and UV protection) contributions to the environments they occupy (Prieto-Barajas et al. Reference Prieto-Barajas, Valencia-Cantero and Santoyo2018). In the McMurdo Dry Valleys, cyanobacteria-dominated microbial mats are the dominant primary producers and strongly influence organic matter budgets (Parker et al. Reference Parker, Simmons, Wharton, Seaburg and Love1982, Burkins et al. Reference Burkins, Virginia, Chamberlain and Wall2000, Lawson et al. Reference Lawson, Doran, Kenig, Des marais and Priscu2004) and N economies (Hawes et al. Reference Hawes, Howard-Williams and Vincent1992, Kohler et al. Reference Kohler, Stanish, Liptzin, Barrett and McKnight2018, Reference Kohler, Singley, Wlostowski and McKnight2023) of the lake, stream and sediment environments they occupy. For example, while stream water chemistry is modulated by upstream glacial-melt conditions and hyporheic interactions that alter the biogeochemistry (Gooseff et al. Reference Gooseff, McKnight, Lyons and Blum2002, Reference Gooseff, McKnight, Runkel and Vaughn2003, Maurice et al. Reference Maurice, McKnight, Leff, Fulghum and Gooseff2002), microbial mat communities also influence water chemistry through biological uptake, recycling and downstream export of N and P (Gooseff et al. Reference Gooseff, McKnight, Runkel and Duff2004, McKnight et al. Reference McKnight, Runkel and Tate2004, Kohler et al. Reference Kohler, Stanish, Liptzin, Barrett and McKnight2018).
Map of eastern Taylor Valley with 16 field sites indicated by open black circles. Inset on bottom right identifies the McMurdo Dry Valleys, Antarctica, with a red square. Ross Sea, Taylor II, Taylor III/IVa and Taylor IVb drifts are delineated by colour, modified from Bockheim et al. (Reference Bockheim, Prentice and McLeod2008). Esri basemap imagery © Earthstar Geographics.
Most studies of microbial mats in the McMurdo Dry Valleys have been conducted in or adjacent to perennial aquatic landscape features (i.e. streams, ponds and lakes; Moorhead et al. Reference Moorhead, Barrett, Virginia, Wall and Porazinska2003, Gooseff et al. Reference Gooseff, McKnight, Runkel and Duff2004, McKnight et al. Reference McKnight, Tate, Andrews, Niyogi, Cozzetto and Welch2007, Stone et al. Reference Stone, Devlin, Hawes, Welch, Gooseff and Takacs-Vesbach2024). Intermittently wet soils and ephemeral, un-channelized streams have received comparatively less attention, even though these landscapes occupy large proportions of the area, and the functioning of microbial mats and biocrusts is relevant to ongoing questions regarding the sources of energy fuelling the simple soil food webs and the regional C balance (e.g. Burkins et al. Reference Burkins, Virginia, Chamberlain and Wall2000, Barrett et al. Reference Barrett, Virginia, Parsons and Wall2006a, Power et al. Reference Power, Thomas, Salvatore and Barrett2024b,Reference Power, Salvatore, Sokol, Stanish, Borges, Adams and Barrettc). Moreover, strong geochemical variation in inorganic P (e.g. Barrett et al. Reference Barrett, Virginia, Lyons, Mcknight, Priscu and Doran2007, Welch et al. Reference Welch, Lyons, Whisner, Gardner, Gooseff, McKnight and Priscu2010) may influence the distribution and activity of microbial mats by controlling their N-fixation potential.
The objective of our study was to investigate the relationships between underlying soil chemistry and microbial mat distribution, composition and function in Taylor Valley, Antarctica. We expected to observe a feedback relationship whereby soil geochemistry (i.e. variation in inorganic P) influences microbial mat occurrence, and in turn the density of microbial mats influences the underlying soil geochemistry. Specifically, we characterized the taxonomic composition, underlying soil geochemistry and pigment composition of microbial mats distributed across two glacial drifts in the Taylor Valley, and we further examined the influence of variable soil inorganic P content on microbial mats and their N-fixation potential, as well as the influence of microbial mats on the C and N budgets of the underlying soils.
Materials and methods
Site description
The McMurdo Dry Valleys (Fig. 1) are the largest contiguous ice-free area on the Antarctic continent, with ~4500 km2 of exposed soil, stream and lake ecosystems (Levy Reference Levy2013). They are one of Earth’s coldest and driest deserts, with mean annual temperatures between -15°C and -30°C and less than 50 mm water equivalent of snowfall annually (Fountain et al. Reference Fountain, Nylen, Monaghan, Basagic and Bromwich2010, Obryk et al. Reference Obryk, Doran, Fountain, Myers and McKay2020). Glacial meltwater during the summer (i.e. 24 h of daylight) feeds streams that flow for an average of 4–9 weeks per year (Wlostowski et al. Reference Wlostowski, Gooseff, McKnight, Jaros and Lyons2016). These streams drain into perennially ice-covered lakes along the floor of most of the valleys, supporting microbial mats within and along the margins of the streams and lakes (McKnight et al. Reference McKnight, Niyogi, Alger, Bomblies, Conovitz and Tate1999). Additionally, other sources of moisture (e.g. snow patches, wetlands and groundwater seeps; Levy et al. Reference Levy, Andrews, Guller, Johnson, King and Pfaff2024) are also activated during the summer, and many of these wetted environments also support microbial mats (Gooseff et al. Reference Gooseff, Barrett and Levy2013).
The parent material of soils in the region are composed of glacial drift deposits (i.e. unsorted rock/soil material deposited from glacial ice or meltwater). In Taylor Valley, where this study was conducted, the soils form from three types of glaciations: 1) advances from the grounded Ross Sea Ice Sheet from the east, 2) advances of glaciers from the East Antarctic Ice Sheet to the west and 3) advances of alpine glaciers along the Taylor Valley walls (Bockheim et al. Reference Bockheim, Prentice and McLeod2008). These glaciations are thought to have occurred over the past 10,000 to 10 million years (Bockheim et al. Reference Bockheim, Prentice and McLeod2008), resulting in very different exposure ages and, given the regional variation in bedrock composition and mineralogy, distinct clast compositions (Denton et al. Reference Denton, Bockheim, Wilson and Stuiver1989, Bate et al. Reference Bate, Barrett, Poage and Virginia2008, Bockheim et al. Reference Bockheim, Prentice and McLeod2008). The Ross Sea drift is on the eastern side of Taylor Valley (Fig. 1) and was deposited more recently than the Taylor drifts on the western side of Taylor Valley. Soil P content is most abundant on the Ross Sea drift due to the weathering of apatite, which is more abundant on these drifts relative to the Taylor Valley sequence (Barrett et al. Reference Barrett, Virginia, Lyons, Mcknight, Priscu and Doran2007, Bate et al. Reference Bate, Barrett, Poage and Virginia2008, Heindel et al. Reference Heindel, Spickard and Virginia2017, Reference Heindel, Lyons, Welch, Spickard and Virginia2018). The geochemistry of the drifts in Taylor Valley has been previously described (e.g. Barrett et al. Reference Barrett, Virginia, Lyons, Mcknight, Priscu and Doran2007, Bate et al. Reference Bate, Barrett, Poage and Virginia2008); however, the impacts of the variation in soil geochemistry on microbial mat composition and activity have not been studied. Additionally, while the influence of microbial mats on stream chemistry has been well documented (e.g. McKnight et al. Reference McKnight, Runkel and Tate2004, Kohler et al. Reference Kohler, Stanish, Liptzin, Barrett and McKnight2018), the influence of microbial mats, due to their N- and C-fixation potential in particular, on underlying soils alongside stream margins is less well understood.
Studies of microbial mats in the McMurdo Dry Valleys have described them primarily by their visible colour, along with the hydrological conditions of their habitat (Alger et al. Reference Alger, McKnight, Spalding, Tate, Shupe and Welch1997, McKnight et al. Reference McKnight, Niyogi, Alger, Bomblies, Conovitz and Tate1999, Stanish et al. Reference Stanish, Nemergut and McKnight2011, Kohler et al. Reference Kohler, Stanish, Crisp, Koch, Liptzin, Baeseman and McKnight2015). A useful typology has emerged from these studies that is based on dominant pigments and visible colour: black, orange, green and red mats (Alger et al. Reference Alger, McKnight, Spalding, Tate, Shupe and Welch1997). The black mats primarily occupy intermittently saturated areas at the margins of lentic and lotic landscapes and beside melting snow patches, whereas orange, green and red mats primarily occupy seasonally inundated areas generally within the thalweg of stream channels (Howard-Williams et al. Reference Howard-Williams, Vincent, Broady and Vincent1986, Alger et al. Reference Alger, McKnight, Spalding, Tate, Shupe and Welch1997, Kohler et al. Reference Kohler, Stanish, Crisp, Koch, Liptzin, Baeseman and McKnight2015). While this colour typology should be understood as a description of community type rather than specific taxa, the generalization that black mats are dominated by Nostoc, green mats by chlorophytes and orange and red mats by Oscillatoria is supported by early microscopy (e.g. Parker et al. Reference Parker, Simmons, Wharton, Seaburg and Love1982, Vincent & Howard-Williams Reference Vincent and Howard-Williams1986, Alger et al. Reference Alger, McKnight, Spalding, Tate, Shupe and Welch1997) and later molecular sequencing approaches (e.g. Van Horn et al. Reference Van Horn, Wolf, Colman, Jiang, Kohler and McKnight2016).
Field sampling
In December 2019 and January 2020, we sampled microbial mats and underlying soils at 16 sites located on the Ross Sea (n = 8) and Taylor drifts (n = 8) of Taylor Valley, Antarctica (Fig. 1 & Table S1). These sites included landscapes that were intermittently wet environments, such as alongside stream channels, pond margins and snowmelt patches. Sampling locations were often within 1 m of water sources but not within standing water. We did not select sites within the thalweg of incised stream channels or in the inundated areas of lakes and ponds as our focus was on the contribution of microbial mats to terrestrial soils. Stream-to-soil ecotones are not always clear in this region (e.g. Harris et al. Reference Harris, Carey, Berry Lyons, Welch and Fountain2007, Gooseff et al. Reference Gooseff, Barrett and Levy2013, Wlostowski et al. Reference Wlostowski, Schulte, Adams, Ball, Esposito and Gooseff2019), but we sought landscape surfaces that were not subject to channelized stream flow, where lotic processes dominate the biogeochemistry and community composition of microbial mats (Gooseff et al. Reference Gooseff, McKnight, Runkel and Duff2004, Kohler et al. Reference Kohler, Stanish, Crisp, Koch, Liptzin, Baeseman and McKnight2015, Zoumplis et al. Reference Zoumplis, Kolody, Kaul, Zheng, Venepally and McKnight2023). Dominant colour typology composition (orange mats, typically described as dominated by Oscillatoria, or black mats, typically considered Nostoc-dominated) was noted for each location (e.g. Alger et al. Reference Alger, McKnight, Spalding, Tate, Shupe and Welch1997).
We identified a 0.5 × 0.5 m area of microbial mat at each site (e.g. Fig. 2) and sampled five plugs of mat material (from the centre and the four corners) using a sterilized #13 brass cork borer (2.27 cm2) for organic matter content, pigment analysis and molecular analysis separately, for a total of 15 plugs. Directly beneath each sampled microbial mat we also collected the underlying soil (0–5 cm depth) using a sterilized cork borer for later geochemical and molecular analysis. Geochemical comparisons to existing, published soil data from the same drift sequences in the vicinity of our sampling sites were conducted (see ‘Discussion’ section), as no bare soils (i.e. soils without overlaying microbial mats) were collected during this field sampling campaign. The five microbial mat and underlying soil samples were composited into one soil and one microbial mat sample per site per analysis. All samples were kept frozen at -20°C in complete darkness until subsequent analyses were performed.
Photographs of a. 0.5 × 0.5 m plot of a dense black microbial mat alongside Crescent Stream (R07, Ross Sea drift), b. a dense black microbial mat alongside an unnamed ephemeral stream near Taylor Glacier (T15, Taylor drifts) and c. a dense orange microbial mat alongside an unnamed stream near Taylor Glacier (T14, Taylor drifts). Photographs taken by E.D. Osburn and S.N. Power.
Soil geochemical analyses
Gravimetric water content (GWC) was measured for each of the soil samples, determined as the mass of water lost after oven drying the soils at 105°C for 24 h. Soil pH and electrical conductivity (EC) were measured using 1:2 and 1:5 soil:deionized H2O slurries, respectively, using a YSI pH meter (YSI, Yellow Springs, OH, USA) and Orion Star conductivity meter (Thermo Fisher Scientific, Waltham, MA, USA). We also extracted inorganic N (NH4 + and NO3 -) in 2 M potassium chloride and inorganic P (PO4 3-) in 0.5 M sodium bicarbonate. We interpret this extractable (soluble) nitrate, ammonium and phosphate as the biologically available N and P in the soil. Major anions were extracted using 1:5 dilutions of soil:deionized H2O. Inorganic N and P were measured on extracts using a Lachat QuikChem flow injection analyser (Hach Company, Loveland, CO, USA; Prokopy Reference Prokopy1995, Knepel Reference Knepel2003). Major anion concentrations were analysed on a Dionex Ion Chromatograph (Thermo Fisher Scientific, Waltham, MA, USA; Pfaff Reference Pfaff1993); only sulphate (SO4 2-) and chloride (Cl-) had consistent concentrations above minimum detection levels. Additionally, we measured soil organic carbon (SOC) and total nitrogen (TN) using an Elementar Vario MAX Cube analyser (Elementar Americas, Inc., Mount Laurel, NJ, USA) after fumigating samples with concentrated hydrochloric acid to remove the influence of carbonates on SOC values (Walthert et al. Reference Walthert, Graf, Kammer, Luster, Pezzotta, Zimmermann and Hagedorn2010).
Microbial mat organic matter and pigment analyses
Each microbial mat sample was measured for organic matter content as ash-free dry mass (AFDM) by oven drying at 100°C for 24 h, weighing the dry mass, combusting at 550°C for 15 h using a muffle furnace, weighing the ashed mass after cooling in a desiccator and estimating AFDM as the mass lost during ashing per mass of dry material (Dean Reference Dean1974, Heiri et al. Reference Heiri, Lotter and Lemcke2001, Santisteban et al. Reference Santisteban, Mediavilla, López-Pamo, Dabrio, Zapata and García2004). Additionally, pigment composition and concentration (Table S2) were estimated at the Paerl Lab, University of North Carolina, using high-performance liquid chromatography (HPLC; Mantoura & Llewellyn Reference Mantoura and Llewellyn1983, Van Heukelem Reference Van Heukelem, Lewitus, Kana and Craft1992, Reference Van Heukelem, Lewitus, Kana and Craft1994, Jeffrey et al. Reference Jeffrey, Mantoura and Wright1997). Pigments were extracted from the microbial mat samples in 100% acetone for 24 h before analysis on the Shimadzu HPLC system (Shimadzu Scientific Instruments, Columbia, MD, USA). The HPLC separated the pigments and measured the absorbance of the extracts by scanning 340–700 nm every 1.28 s. These data were analysed using Shimadzu’s LabSolutions Lite software and a combination of peak retention time, absorbance spectrum shape/signature, maximum wavelength and the similarity match of the unknown pigment to a standard to identify the various pigments (Jeffrey et al. Reference Jeffrey, Mantoura and Wright1997). Pigments were then quantified from their peak areas, calculated at 388 nm for scytonemin and reduced scytonemin and 440 nm for all other pigments. A multipoint calibration curve was generated by injecting volumes of known quantities of pure pigment standards (DHI, Denmark) and then calculating the peak areas of those pigments. The standards’ peak areas were then used to calculate the slopes (response factor) for all pigments, which were multiplied by the samples’ peak areas of the chromatogram at 388 or 440 nm to quantify the pigments extracted in μg g-1 dry mat.
DNA extraction and quantitative polymerase chain reaction
We determined the relative abundance of all nitrogenase (nif) genes, which are indicators of N-fixation potential, using shotgun metagenome sequencing and the absolute abundance of nifH specifically using quantitative polymerase chain reaction (qPCR) for both the microbial mats and underlying soils. DNA was extracted from ~1.5 g soil and ~ 0.5–1.0 g microbial mat material using the DNeasy PowerSoil kit (QIAGEN), and the extracts were quantified using a Qubit 2.0 fluorometer (Thermo Fisher Scientific). For qPCR, we used the IGK3/DVV primer set (Ando et al. Reference Ando, Goto, Meunchang, Thongra-ar, Fujiwara, Hayashi and Yoneyama2005) because it exhibits less bias than other nifH primer sets (Gaby & Buckley Reference Gaby and Buckley2017). Samples were amplified in triplicate and PCR reactions contained 10 μl Quantitect SYBR Green qPCR Mastermix (QIAGEN), 3 μl of 10 μM working stocks of each of the forward and reverse primers (for final primer concentrations of 1.5 μM) and 4 μl of DNA template for a total reaction volume of 20 μl. The high primer concentration we used was intended to maximize the efficiency of the qPCR reactions (Gaby & Buckley Reference Gaby and Buckley2017). Thermal cycling conditions were as follows: 15 min at 95°C followed by 40 cycles of 15 s at 95°C, 30 s at 56°C and 30 s at 72°C. Standard curves were generated by amplifying serial dilutions of the target regions, with amplification efficiencies of 84.5% for the microbial mat samples and 74.2% for the soil samples, and both R 2 values > 0.99. For the qPCR gene standard, we used the nifH gene sequence from Nostoc punctiforme strain PCC 73102, which has a high similarity (99%) to Antarctic Nostoc sequences (Jungblut & Neilan Reference Jungblut and Neilan2010). Amplification specificity was assessed using melt curve analysis. The nifH gene copy numbers were corrected for dry soil mass and wet mat mass.
Metagenomic sequencing and taxonomic assignment
DNA libraries were prepared for shotgun metagenome sequencing using a KAPA HyperPrep Kit (Roche Molecular Systems, Inc.). Libraries were sequenced on an Illumina NextSeq500 in a 150 bp paired-end sequencing run at the Duke University School of Medicine Sequencing and Genomic Technologies Shared Resource. Raw sequence reads were uploaded to MG-RAST (accession# mgp95386) for merging of paired ends, quality control and annotation using the recommended MG-RAST default parameters (Meyer et al. Reference Meyer, Paarmann, D'Souza, Olson, Glass and Kubal2008). Functional annotations were performed by aligning the metagenomic sequences to the SEED Subsystems database (Overbeek et al. Reference Overbeek, Olson, Pusch, Olsen, Davis and Disz2014), while taxonomic annotations were performed using the RefSeq and SILVA databases. For all functional and taxonomic annotations, we used an e-value cutoff of 1e-5, a sequence similarity threshold of 80% and a minimum alignment length of 20 bp. After processing with MG-RAST, we retained an average of ~7.2 million sequences per sample, ranging from ~4.3 million to ~8.9 million sequences per sample. Of these sequences, an average of 3.91% failed quality control, 1.10% were rRNA genes, 41.25% were annotated proteins and 57.65% were unknown proteins. To quantify N-fixation gene relative abundances, we summed all sequences annotated as nif genes for each sample and expressed nif gene relative abundance as the percentage of all annotated reads.
Taxonomy was assigned to the processed reads using the NCBI RefSeq and SILVA (version 138.1) databases. We determined community composition by analysing the relative abundance of prokaryotic taxa that classified to at least a phylum (bacterial or archaeal). All eukaryotic taxa and any taxa unclassified to a phylum were removed from the analysis. To account for differences in sequence depth among samples, we used rarefied sequence counts. The rarefaction curves for the RefSeq data saturated, whereas the ones for the SILVA data did not; thus, we based our interpretations and conclusions primarily on the RefSeq results.
Statistical analyses
All statistical analyses were performed in R Statistical Software (R Core Team 2021). Figures were created in R using the ggplot2 package (Wickham Reference Wickham2016). One-way analysis of variance (ANOVA; aov function, stats package; R Core Team 2021) was used to determine differences in soil geochemistry across drift types. Relative abundance of prokaryotic genera was determined using the following packages: vegan for the rarefaction curve analysis (Oksanen et al. Reference Oksanen, Simpson, Blanchet, Kindt, Legendre and Minchin2022), funrar for relative abundance calculation (Grenié et al. Reference Grenié, Denelle, Tucker, Munoz and Violle2017) and reshape, tidyr, plyr and stringr for data manipulation before visualizing the relative abundance (Wickham Reference Wickham2007, Reference Wickham2011, Reference Wickham2023, Wickham et al. Reference Wickham, Vaughan and Girlich2023). Diversity was assessed using alpha (Shannon) and beta (Bray-Curtis dissimilarity) diversity metrics with the diversity function in the vegan package. A one-way ANOVA was used to determine differences in Shannon diversity by mat type. Bray-Curtis dissimilarity was calculated using the vegdist function in the vegan package after converting the data to relative abundances. A non-metric multidimensional scaling (NMDS) ordination was created with permutational analysis of variance (PERMANOVA) to visualize and analyse differences in prokaryotic community composition between mat type (metaMDS and adonis2 functions, vegan package).
One-way ANOVAs were also performed to assess differences in AFDM and pigment concentrations by mat type. We used principal component analysis (PCA) to visualize the relative abundance of pigments across sites (princomp function, stats package; envfit function, vegan package), and we used PERMANOVA to assess mat type differences with Euclidean distance matrices. Additionally, the dispersion of the pigment relative abundance was calculated by mat type (betadisper function, vegan package), and a one-way ANOVA was used to assess the significance of the difference in dispersion by mat type. The relationships between metagenomic-derived nifH relative abundance and the qPCR-derived number of nifH genes for soils and microbial mats were assessed using Pearson correlation coefficients (r). Differences in nif and nifH gene abundance for the soils and microbial mats were assessed by mat type using one-way ANOVAs.
Mean physicochemical variables for soils underlying microbial mats averaged by glacial drift composition: Ross Sea drift (n = 8) and Taylor drifts (n = 8) with ± 1 standard deviation. One-way analysis of variance tests were run between Ross Sea and Taylor drift soils for each parameter, and superscript letters represent differences (measurements with dissimilar letters were significantly different from each other, P < 0.05).
EC = electrical conductivity; GWC = gravimetric water content; SOC = soil organic carbon; TN = total nitrogen.
We applied polynomial regressions between AFDM and SOC, TN and the relative abundance of nif in soils. Assumptions of linear and polynomial regression were assessed, including normality (Q-Q plot assessment, Shapiro-Wilk normality test; shapiro.test function, stats package), homoscedasticity (Breusch-Pagan test; bptest function, lmtest package; Zeileis & Hothorn Reference Zeileis and Hothorn2002) and no autocorrelation (Durbin-Watson test; dwtest function, lmtest package), and the data met all assumptions for polynomial regression. Spearman correlation coefficients (ρ) were calculated to assess relationships between the inorganic P of the underlying soils and microbial mat parameters including AFDM, relative abundance of nif and pigment concentrations of chlorophyll-a, scytonemin, β-carotene and canthaxanthin, as the distribution of these data did not meet the assumptions required for parametric analyses. We also used PCA to visualize variation in N and P availability, AFDM, nif gene relative abundance and scytonemin concentration among sites, and we assessed differences by mat type using PERMANOVA with Euclidean distance matrices. Additionally, the dispersion of these results was calculated by mat type (betadisper function, vegan package), and a one-way ANOVA was used to assess the significance of the difference in dispersion by mat type. For all statistical analyses, P < 0.05 was considered statistically significant.
Results
Distribution of ‘black’ and ‘orange’ microbial mats
Dense black microbial mats were the most conspicuous mat type in the intermittently wet environments we sampled on the Ross Sea drift, where we encountered dense black mats at the margins of streams and downhill of seasonal snowpacks (Fig. 2 & Table S1). Orange mats are abundant in the thalweg of the streams in this part of Taylor Valley, as has been previously described (Howard-Williams et al. Reference Howard-Williams, Vincent, Broady and Vincent1986, Alger et al. Reference Alger, McKnight, Spalding, Tate, Shupe and Welch1997, Kohler et al. Reference Kohler, Stanish, Crisp, Koch, Liptzin, Baeseman and McKnight2015), but they were not apparent in the drier ‘terrestrial’ environments we examined in this study. The Taylor drifts had a lower abundance of microbial mats overall in comparison to the Ross Sea drift. Unlike the Ross Sea drift, we were able to identify and sample orange mat communities on the Taylor drifts outside of the thalweg of streams, typically on hillslopes downhill from large snowpacks that feed ephemeral wetland landscapes (e.g. landscapes that do not support flowing surface water in most years, such as the Wormherder Creek wetland; Harris et al. Reference Harris, Carey, Berry Lyons, Welch and Fountain2007, Wlostowski et al. Reference Wlostowski, Schulte, Adams, Ball, Esposito and Gooseff2019). Black mats were less common on the Taylor drifts but were found along ephemeral stream margins. We sought to collect an equal distribution of black and orange mats across the drift types; however, the observed occurrences of mat types resulted in an asymmetrical distribution of mat types across the drift sequences.
Soil geochemical characterization below microbial mats
Soil geochemical characteristics below microbial mats varied among the plots, mainly associated with differences between the Ross Sea and Taylor drifts. Soil pH, inorganic P, SOC and TN were significantly higher in samples collected from the Ross Sea compared to the Taylor drifts (ANOVA, P < 0.05; Table I). For example, Ross Sea drift soil contained four times more biologically available inorganic P compared to the Taylor drifts, with an average of 2.36 μg P g-1 dry soil compared to 0.60 μg P g-1 dry soil, and four times lower N:P ratios than the Taylor drift samples. The average extractable, inorganic N:P ratios (by mass) of the Ross Sea and Taylor drift samples were 0.87 and 3.78, respectively. SOC was two times higher in the Ross Sea drift samples compared to the Taylor drifts, with an average of 1.79 mg C g-1 dry soil compared to 0.85 mg C g-1 dry soil. TN was also two times higher in the Ross Sea drift samples compared to the Taylor drifts, with an average of 0.19 mg N g-1 dry soil compared to 0.11 mg N g-1 dry soil. Soil GWC, EC, NH4 +, NO3 -, SO4 2- and Cl- were not significantly different between the Ross Sea and Taylor drift samples (Table I).
Community composition of microbial mats
Community composition varied among the microbial mat samples, mainly associated with differences between black and orange mats. Taxonomy assigned using the RefSeq and SILVA databases generated similar results, with only small differences in relative abundance at the phylum level (primarily for Cyanobacteria, Proteobacteria and Bacteroidetes; Figs 3a & S1a), and no meaningful differences at the genus level, which were dominated by cyanobacteria taxa (Figs 3b & S1b). At the phylum level, all microbial mat types sampled consisted primarily of a mix of Cyanobacteria, Proteobacteria and Bacteroidetes (Fig. 3a). In black mats of both the Ross Sea and Taylor drifts, Proteobacteria was the most abundant phylum, followed by Cyanobacteria and Bacteroidetes. In contrast, orange mats were dominated by Cyanobacteria at the phylum level. At the genus level, every microbial mat sample contained a large proportion of rare genera (i.e. genera constituting less than 1.4% of sequences), contributing to more than 50% of the relative abundance in most of the samples (Fig. 3b). Black mats had a greater proportion of rare genera compared to orange mats. The cyanobacteria in black mats were predominantly Nostoc, whereas in orange mats Oscillatoria was typically the most abundant taxa, followed by Nostoc and Cyanothece. Shannon diversity was significantly greater in black mats, with an average of 5.19 compared to an average Shannon diversity of 4.17 in orange mat communities (ANOVA, P < 0.05). Black and orange mat community compositions were significantly different (PERMANOVA, P = 0.001), and NMDS visualization using Bray-Curtis dissimilarities exhibited significant differences in taxonomic composition between the mat types (Fig. 4).
Relative abundances of a. prokaryotic phyla and b. prokaryotic genera identified within each of the 16 samples using the NCBI RefSeq sequence library. The remaining prokaryotic taxa of less than 1.4% are represented in ‘Other’. Samples collected from the Ross Sea drift begin with ‘R’ (i.e. R01–R08), and samples collected from the Taylor drifts begin with ‘T’ (i.e. T09–T16). The x-axis label colours represent whether the samples were collected from black or orange microbial mats based upon visual field surveys.
Non-metric multidimensional scaling (NMDS) ordination of the community composition (by genera) of each mat type using Bray-Curtis dissimilarities. The shown P-value is the mat type effect from permutational analysis of variance.
Organic matter content and pigment characterization of microbial mats
Organic matter content as AFDM and pigment concentration varied among the samples, mainly associated with differences in black and orange microbial mats. AFDM ranged over an order of magnitude from 12.7 to 185.8 mg g-1 dry mat and was significantly greater in black mats than in orange mats (ANOVA, P < 0.05; Table II). Black microbial mats contained three times the mass of organic matter than orange mats, with an average of 78.5 mg AFDM g-1 dry mat.
Average ash-free dry mass (AFDM) and pigment concentrations averaged by microbial mat type: black mats (n = 11) and orange mats (n = 5) with ± 1 standard deviation. One-way analysis of variance tests were run between black and orange mats for each parameter, and superscript letters represent differences (measurements with dissimilar letters were significantly different from each other, P < 0.05).
B-Car = β-carotene; Cantha = canthaxanthin; Chl-a = chlorophyll-a; Chl-b = chlorophyll-b; Echine = echinenone; Myxo = myxoxanthophyll; Scyto = scytonemin; Scyto-Red = reduced scytonemin; Zea = zeaxanthin.
The microbial mats sampled contained photosynthetic and accessory pigments common to many cyanobacteria: chlorophylls, multiple carotenoids and extracellular sheath pigments. Out of 23 pigments analysed using HPLC (Table S2), nine pigments had detectable concentrations present within the sampled microbial mats (Table II). Black microbial mats had 68 times the total pigment concentration (267.5 μg g-1 dry mat on average) of orange mats (3.9 μg g-1 dry mat on average; ANOVA, P < 0.001; Table II). Scytonemin, reduced scytonemin, myxoxanthophyll and β-carotene concentrations were significantly different between black and orange mats (ANOVA, P < 0.05; Table II). Black mat pigment composition was dominated by scytonemin, which totalled > 83% of the total pigment relative abundance on average. Black mats contained 165 times the concentration of scytonemin compared to orange mats, with an average of 225.3 μg g-1 dry mat. Similarly, black mats also contained greater concentrations of reduced scytonemin: 35 times the amount measured in orange mats. Black mats contained six times the concentration of β-carotene than orange mats; however, β-carotene accounted for a greater relative abundance of the total pigment abundance of orange mats. Similarly, black mats contained, on average, a greater concentration of chlorophyll-a, whereas the relative abundance of chlorophyll-a was greater in orange mats. Multivariate visualization (PCA) of pigment relative abundance illustrates statistically significant differences among black and orange microbial mat communities based on microbial mat type (PERMANOVA, P = 0.001), with orange mats containing a greater variation in pigment type (ANOVA on dispersion, P < 0.01; Fig. 5). For example, three of the orange mat samples contained myxoxanthophyll (T12, T13, T14), whereas none of the black mat samples did.
Principal component (PC) analysis ordination on correlation of the relative abundances of the pigments identified in the 16 microbial mat samples, where ‘scyto’ is scytonemin, ‘myxo’ is myxoxanthophyll, ‘zea’ is zeaxanthin, ‘chl.a’ is chlorophyll-a, ‘chl.b’ is chlorophyll-b, ‘B.car’ is β-carotene, ‘cantha’ is canthaxanthin and ‘echine’ is echinenone. Samples collected from the Ross Sea drift begin with ‘R’ (i.e. R01–R08), and samples collected from the Taylor drifts begin with ‘T’ (i.e. T09–T16). The label colours represent whether the samples were collected from black or orange microbial mats. Vectors represent correlations of the relative abundance of each pigment with PC analysis ordination axes (all displayed correlations are statistically significant, P < 0.01). The shown P-value is the mat type effect from permutational analysis of variance.
Many of these pigments are photoprotective pigments that are important for microbial communities in the UV-intense McMurdo Dry Valleys, especially for ‘terrestrial’ communities that lack ice or water cover to aid in light absorption (Vincent et al. Reference Vincent, Downes, Castenholz and Howard-Williams1993). Scytonemin is a photoprotective, extracellular sheath pigment (Garcia-Pichel et al. Reference Garcia-Pichel, Sherry and Castenholz1992) that dominates black and orange mats. Differences in pigment composition were associated with differences in taxonomic composition even within black and orange mats. For example, samples R08 and T13 are notable outliers in the ordination of pigment content (Fig. 5). T13 contained the lowest relative abundance of scytonemin (10.7%) and the highest relative abundance of myxoxanthophyll (8.5%). Field observations indicated that R08 was a dense colony of Nostoc with a reddish colour in contrast to the other black mat samples and had the highest concentration of canthaxanthin (2.9 μg g-1 dry mat; 1% relative abundance), an orange-red ketocarotenoid. The taxonomic composition of R08 was also different from most of the other sampled black mats, with less rare taxa present and the highest relative abundance of Nostoc.
Average nitrogen fixation gene abundance for black mats (n = 11) and orange mats (n = 5) with ± 1 standard deviation. Data are separated by those derived from metagenomic methods and quantitative polymerase chain reaction (qPCR) methods. The measures of nitrogen fixation potential are represented by the relative abundance of all nif genes and gene abundance of nifH specifically. One-way analysis of tests were run between black and orange mats for each parameter, and superscript letters represent differences (measurements with dissimilar letters were significantly different from each other, P < 0.05).
Spearman correlation coefficients (ρ) between the inorganic phosphorus concentration of underlying soils and microbial mat parameters. Values in bold indicate statistically significant relationships based on P < 0.05.
AFDM = ash-free dry mass; B-Car = β-carotene; Cantha = canthaxanthin; Chl-a = chlorophyll-a; Scyto = scytonemin.
Nif gene characterization of microbial mats and soils
There was a significant linear correlation between the metagenomic-derived nifH relative abundance and the qPCR-derived number of nifH genes for soils (r = 0.73, P < 0.01) and microbial mats (r = 0.72, P < 0.01). The gene copy number and relative abundance of nif and nifH for mats and underlying soils (from both qPCR and metagenomics methods) were significantly different between black and orange mats (ANOVA, P < 0.05; Table III). Overall, black microbial mats had greater gene copy numbers and relative abundances of nif and nifH genes compared to orange mats.
Polynomial regressions of a. soil organic carbon (SOC), b. total nitrogen (TN) and c. relative abundance of soil nif vs microbial mat organic matter content as ash-free dry mass (AFDM). Symbol colours distinguish black and orange microbial mats, and symbol shapes distinguish on which glacial drift the mats were sampled. The 95% confidence intervals of the regression lines are shown in grey.
Relationships among microbial mats and underlying soil geochemistry
Significant correlations between the microbial mats and the soils underlying them were apparent, particularly in the mat characteristics (i.e. AFDM and pigment content) and the soil geochemistry. Soil inorganic P was significantly correlated with mat AFDM, nif, chlorophyll-a and β-carotene concentrations (ρ = 0.53, 0.58, 0.57 and 0.63, respectively, P < 0.05; Table IV). Additionally, the AFDM of the microbial mats was positively correlated with underlying SOC, TN and the relative abundance of nif in the soils (R 2 = 0.66, P < 0.001; R 2 = 0.57, P < 0.01; R 2 = 0.44, P < 0.05, respectively; Fig. 6). A PCA ordination of N and P availability, AFDM, nif relative abundance and scytonemin concentration exhibits statistically significant separation of microbial mat types (PERMANOVA, P = 0.001), with black mats containing a greater variation in biogeochemical parameters (ANOVA on dispersion, P < 0.05; Fig. 7). This ordination illustrates a strong association between the black mat samples and inorganic P and TN, whereas orange mat samples were less related to variation in inorganic P and TN.
Discussion
Glacial drift composition influences soil chemistry and microbial mats
The landscape history and glacial drift composition of Taylor Valley influences its soil chemistry, which in turn influences its microbial mat distribution, composition and activity. The terrestrial surfaces of the Ross Sea drift in the east and more coastal region have greater P availability than Taylor drifts due to higher apatite in the drift clast composition and greater weathering (Bate et al. Reference Bate, Barrett, Poage and Virginia2008, Heindel et al. Reference Heindel, Spickard and Virginia2017, Reference Heindel, Lyons, Welch, Spickard and Virginia2018). Our data show that the soils underlying the microbial mats on the Ross Sea drift contained significantly more inorganic P and SOC in comparison to the soils underlying the microbial mats on the Taylor drifts (Table I). Inorganic N content was not significantly different between the Ross Sea and Taylor drifts, probably due to the flushing of N (particularly nitrate) in these intermittently wet environments where microbial mats are most encountered. We suggest that the greater content of SOC in soils collected from the Ross Sea drift is associated with greater availability of soil inorganic P, which potentially enhances the capacity of microbial mats to fix N and C. Interpreting these observations through the lens of landscape history provides better understanding of the contemporary variation in microbial mat composition and functioning.
Principal component (PC) analysis ordination on correlation of the N and P availability of underlying soils, organic matter content as ash-free dry mass (AFDM), nif gene relative abundance and scytonemin concentration of the 16 microbial mat samples. Samples collected from the Ross Sea drift begin with ‘R’ (i.e. R01–R08), and samples collected from the Taylor drifts begin with ‘T’ (i.e. T09–T16). The colours represent whether the samples were collected from black or orange microbial mats. Vectors represent correlations of each parameter with PC analysis ordination axes (all displayed correlations are statistically significant, P < 0.05, while NH4 is marginally significant, P < 0.1). The shown P-value is the mat type effect from permutational analysis of variance.
In the intermittently wet environments outside of the stream channels that we sampled, black microbial mats were more common on the Ross Sea drift, whereas orange mats were more common on the Taylor drifts. The utility of this colour typology is validated by our data showing that the prokaryotic community composition was distinct between the two mat types (Figs 3 & 4). Specifically, Nostoc-dominated ‘black’ mats, regardless of underlying glacial drift composition, hosted a complex and diverse community of autotrophic and heterotrophic microbes across many bacterial phyla with a high proportion of rare taxa, whereas the orange mats consisted primarily of cyanobacteria (Oscillatoria) and had lower Shannon diversity values. This clear distinction in microbial diversity and the associated variation in pigment content and composition (Fig. 5) support a simple comparison of the different characteristics of these two mat types and their potential influences on underlying soil properties and processes. For example, black microbial mats contained significantly more organic matter in the form of AFDM, higher absolute and proportional scytonemin contents, as well as higher N-fixation potential, and they appeared to be preferentially located on P-rich soils (Tables II & III). Moreover, inorganic P concentration was significantly correlated with microbial mat organic matter content (AFDM), N-fixation potential and pigment concentration (Table IV).
Although the asymmetrical distribution of black and orange mats across the two drifts prevents conclusive inferences from being drawn regarding the influence of geochemistry on microbial mat distribution and activity, we hypothesize that the greater abundance of P on the Ross Sea drift may select for a greater abundance of active N-fixers, such as Nostoc. Specifically, our data suggest that the distribution of Nostoc black mats is strongly associated with higher inorganic P, possibly because there is a greater advantage of active N fixation when ample P is present. For example, high P availability has been shown to favour N-fixers in freshwater environments outside of Antarctica (Howarth et al. Reference Howarth, Marino and Cole1988, Andersson et al. Reference Andersson, Höglander, Karlsson and Huseby2015, Tierney & Wurzburger Reference Tierney and Wurzburger2024). The greater apparent occurrence of orange mats on the lower-P Taylor drifts may be because they are more competitive in an environment where the strongly N-fixing black mats are at less of an advantage. Although there are orange mats on the Ross Sea drift, they are primarily located within the stream thalweg, where inorganic P is typically limiting and has been weathered out (Bate et al. Reference Bate, Barrett, Poage and Virginia2008, Heindel et al. Reference Heindel, Spickard and Virginia2017, Reference Heindel, Lyons, Welch, Spickard and Virginia2018) and where their isotopic composition suggests that turnover of N from Nostoc further upstream supports their nutrient requirements (Kohler et al. Reference Kohler, Stanish, Liptzin, Barrett and McKnight2018). We sampled exclusively at the stream margins, where we could not locate any orange mats on the Ross Sea drift. We encourage further studies on these terrestrial communities to determine the underlying mechanisms of their distribution along varying inorganic P concentrations. Although a variety of climate, hydrological and geomorphological factors contribute to the abundance, distribution and functioning of microbial mats (McKnight et al. Reference McKnight, Niyogi, Alger, Bomblies, Conovitz and Tate1999, Stanish et al. Reference Stanish, Nemergut and McKnight2011, Kohler et al. Reference Kohler, Stanish, Crisp, Koch, Liptzin, Baeseman and McKnight2015, Levy et al. Reference Levy, Cary, Joy and Lee2020, Power et al. Reference Power, Thomas, Salvatore and Barrett2024b), our work here demonstrates how soil geochemistry and landscape history also influence the prevalence of terrestrial microbial mats.
Microbial mats enrich underlying soil organic matter
While our data show that the geochemical variation in inorganic P along Taylor Valley is a factor influencing microbial mat distribution, microbial mats also make significant contributions to the chemical compositions of the underlying soils, especially to their greater SOC and TN (Fig. 6). Although in this study we did not sample soils from uncolonized areas, previous studies conducted in the same general area and on the same drifts have reported four times less SOC on areas uncolonized by mats (0.43 mg C g-1 dry soil on the Ross Sea drift and 0.19 mg C g-1 dry soil on the Taylor drifts; Barrett et al. Reference Barrett, Virginia, Hopkins, Aislabie, Bargagli and Bockheim2006b) compared to the SOC values we estimated beneath microbial mats (1.79 mg C g-1 dry soil on the Ross Sea drift and 0.85 mg C g-1 dry soil on the Taylor drifts; Table I). Similarly, soils without microbial mats were shown to contain three to nine times less TN (0.02 mg N g-1 dry soil on the Ross Sea drift and 0.03 mg N g-1 dry soil on the Taylor drifts; Barrett et al. Reference Barrett, Virginia, Hopkins, Aislabie, Bargagli and Bockheim2006b) compared to the TN values we estimated here for soils underlying microbial mats (0.19 mg N g-1 dry soil on the Ross Sea drift and 0.11 mg N g-1 dry soil on the Taylor drifts; Table I).
All samples contained a high abundance of nif and nifH genes, suggesting a strong N-fixation potential in both black and orange microbial mats, as has been previously shown primarily for black microbial mats in the McMurdo Dry Valleys (e.g. Kohler et al. Reference Kohler, Stanish, Liptzin, Barrett and McKnight2018, Reference Kohler, Singley, Wlostowski and McKnight2023, Coyne et al. Reference Coyne, Parker, Lee, Sohm, Kalmbach and Gunderson2020). In addition, although all nine of the most abundant genera of the black and orange mats are known to be common N-fixers, both metagenomic approaches and qPCR indicated that the Nostoc-dominated black mats had significantly greater N-fixation potential than the Oscillatoria-dominated orange mats, as has been previously reported (Howard-Williams et al. Reference Howard-Williams, Priscu and Vincent1989, Kohler et al. Reference Kohler, Stanish, Liptzin, Barrett and McKnight2018, Coyne et al. Reference Coyne, Parker, Lee, Sohm, Kalmbach and Gunderson2020). The high N-fixation potential of these mat types probably accounts for the higher TN content in the underlying soils due to N fixation of the overlaying microbial mats.
Conceptual model illustrating the hypothesized causal relationships between glacial drift composition, soil nutrients and organic matter and microbial mats based upon our observations in this study. Arrows indicate significant material fluxes between stocks. Bold arrows indicate the strongly positive relationships described in this study, whereas the dashed arrow indicates a weaker positive relationship.
Aquatic microbial mats influence stream chemistry in Taylor Valley (e.g. McKnight et al. Reference McKnight, Runkel and Tate2004, Kohler et al. Reference Kohler, Stanish, Liptzin, Barrett and McKnight2018) through biological uptake, recycling and export downstream. For example, using stable isotope approaches, Kohler et al. (Reference Kohler, Stanish, Liptzin, Barrett and McKnight2018) showed that black mats were the dominant N-fixers in Von Guerard Stream and that N was repeatedly recycled and assimilated downstream as other sources of N were exhausted. Our data show that terrestrial microbial mats at the margins of streams also influence underlying soil geochemistry. Although the differences in soil inorganic P are of geochemical origin from the weathering of greater apatite content on the Ross Sea drift (Bate et al. Reference Bate, Barrett, Poage and Virginia2008, Heindel et al. Reference Heindel, Spickard and Virginia2017, Reference Heindel, Lyons, Welch, Spickard and Virginia2018), the greater TN found in the soils underlying microbial mats is probably of biotic origin from N fixation by the cyanobacterial mats.
Differences in N-fixation potential may also explain the microbial responses to physical stressors in the Antarctic environment, such as UV radiation. For example, Fleming & Castenholz (Reference Fleming and Castenholz2008) reported a connection between scytonemin synthesis and N sources and availability. In a laboratory experiment they showed that Nostoc synthesized significantly more scytonemin while fixing N than when utilizing available nitrate or ammonium (Fleming & Castenholz Reference Fleming and Castenholz2008). Fleming & Castenholz (Reference Fleming and Castenholz2008) hypothesized that a restriction in N availability could cause an increase in UV radiation sensitivity in Nostoc, which therefore signals the increase in scytonemin synthesis. In our field-based studies, we found a significant, positive correlation between scytonemin concentration and nif relative abundance as N-fixation potential (r = 0.56, P < 0.05; Fig. 7). Although a mechanism for this remains unexamined, we hypothesize that this relationship may also be associated with greater UV stress under conditions of N limitation. The McMurdo Dry Valleys experience exceptionally high UV radiation, and it is possible that the increase in scytonemin with mats that have a higher N-fixation potential is due to greater photosensitivity. Alternatively, scytonemin may be produced in larger quantities to protect the nitrogenase enzyme complex, which has been shown to be sensitive to UV radiation (Kumar et al. Reference Kumar, Tyagi, Jha, Srinivas and Singh2003, Singh et al. Reference Singh, Jha, Rana, Mishra, Kumari and Singh2023).
Our work here and previous studies support the notion that microbial mats impart an ‘island of fertility’ effect (sensu Schlesinger et al. Reference Schlesinger, Reynolds, Cunningham, Huenneke, Jarrell, Virginia and Whitford1990) on local soils and sediments through their influence on organic matter availability and N availability (Hawes et al. Reference Hawes, Howard-Williams and Vincent1992) and through promoting diverse soil fauna communities (e.g. nematodes, tardigrades, rotifers and ciliates; Simmons et al. Reference Simmons, Wall, Adams, Ayres, Barrett and Virginia2009). Specifically, our results demonstrate that terrestrial microbial mats are composed of diverse organisms that fix both C and N and contribute leached organic matter and nutrients into underlying soils. Previous studies have shown that the same holds true for aquatic microbial mats contributing organic matter and nutrients into stream waters as well, thereby supporting diverse soil and sediment communities (Moorhead et al. Reference Moorhead, Barrett, Virginia, Wall and Porazinska2003, Simmons et al. Reference Simmons, Wall, Adams, Ayres, Barrett and Virginia2009, Stanish et al. Reference Stanish, Nemergut and McKnight2011, Kohler et al. Reference Kohler, Stanish, Liptzin, Barrett and McKnight2018). Moreover, previous work has also shown that microbial mats have a homogenizing effect on the underlying soil bacterial communities in this region (i.e. bacterial communities below microbial mats support similar taxonomic compositions regardless of glacial drift composition or overlaying microbial mat type; Risteca Reference Risteca2023). In addition to influencing soil nutrient cycles and community composition, microbial mats can also influence soil hydrology due to their low permeability, which promotes higher moisture content in the soils below them (Perillo et al. Reference Perillo, Maisano, Martinez, Quijada and Cuadrado2019) and provides shelter from wind and UV radiation. These are essential functions, especially in the harsh and nutrient-poor terrestrial environment of the McMurdo Dry Valleys. Current (Nielsen et al. Reference Nielsen, Katurji, Zawar-Reza and Meyer2023, Reference Nielsen, Katurji, Zawar-Reza and Cullen2024, Barrett et al. Reference Barrett, Adams, Doran, Dugan, Myers and Salvatore2024) and anticipated (Vignon et al. Reference Vignon, Roussel, Gorodetskaya, Genthon and Berne2021) changes in climate and hydrology will have unknown influences on the distribution and activity of microbial mats in the McMurdo Dry Valleys. For example, greater snowfall or even rain in a potentially warmer climate regime would probably expand microbial mat habitats to currently arid and inactive soils, whereas increased föhn winds and associated drier conditions could decrease low-density microbial mats in currently marginal habitats. Any future changes in microbial mat distribution and activity will probably have marked effects on underlying soil biota and chemistry.
Conclusion
Here, we show how geochemical variation in inorganic P influences the occurrence, composition and ecological functioning of microbial mats, the dominant primary producers of many extreme ecosystems. The ‘terrestrial’ microbial mats we examined in this study enhance underlying soil C and N cycling and provide important sources of the energy and N fuelling soil food webs. Thus, understanding the distribution and activity of terrestrial microbial mats is essential to the regional C balance. This work highlights the recursive nature of the relationships between microbial mats and the soils underlying them and presents a conceptual model (Fig. 8) that could be useful for anticipating how changes in climate and regional weather events could alter microbial mat distribution and activity, and thereby the soil food webs and overall organic matter budgets in this region.
Supplementary material
To view supplementary material for this article, please visit http://doi.org/10.1017/S0954102025000094.
Acknowledgements
We acknowledge Antarctic Support Contractors and Air Center Helicopters for providing essential operational support in the field. We also thank Bobbie Niederlehner (Virginia Tech) for her assistance with analytical chemistry techniques and Karen Rossignol (Paerl Lab, University of North Carolina at Chapel Hill Institute of Marine Sciences) for leading the HPLC analysis on the pigment samples. Additionally, we thank Dr Cayelan Carey and Dr Erin Hotchkiss for providing invaluable feedback during the development of this manuscript. Lastly, we thank Dr Joseph Levy and an anonymous reviewer whose thoughtful feedback improved this manuscript.
Data availability
All data presented in this study are archived in the Environmental Data Initiative (EDI) repository (Power et al. Reference Power, Osburn and Barrett2024a). Metagenomic sequences are available on the MG-RAST database under project ID mgp95386: https://www.mg-rast.org/linkin.cgi?project=mgp95386.
Financial support
This work was supported by the National Science Foundation (NSF) through grants #1637708 and #2224760 to the McMurdo Dry Valleys Long Term Ecological Research Project.
Competing interests
The authors declare that they have no competing interests.
Ethical standards
Permission to enter the Canada Glacier Antarctic Specially Protected Area (ASPA 131) and sample soils and microbial mats was given by the National Science Foundation Division of Polar Programs through Antarctic Conservation Act Permit.
Author contributions
All authors contributed to the design of the study and the field sampling. SNP and EDO conducted sample preparation and laboratory analyses. SNP led the statistical analyses and interpretation with support from EDO and JEB. All authors contributed to the writing and editing of the manuscript.
Open access
