Introduction

Many microbial metabolisms affect the concentrations of soluble gases. For example, O 2 concentrations are locally very high during oxygenic photosynthesis, whereas sulfate reduction and methanogenesis can increase concentrations of H 2 S and CH 4 in anoxic settings (Biddanda et al., 2015;Nold et al., 2013;Voorhies et al., 2012). H 2 S, CH 4 , and O 2 -rich bubbles can nucleate in microbial mats when gas production causes supersaturation of dissolved gases in the local environment (Bosak et al., 2010;Gerdes et al., 2000;Voorhies et al., 2012). Environmental concentrations of dissolved gases, hydrostatic pressure, temperature, water chemistry, and rates of microbial production or consumption of gases all influence bubble nucleation and growth. Bubbles can be stabilized in benthic communities when they form cohesive mats. Thus, both ecological processes and environments influence the dynamics of bubble formation.

RESEARCH ARTICLE

10.1029/2024JG008516

Key Points:

• Benthic microbial mats in gas-saturated zones of ice-covered lakes in the McMurdo Dry Valleys have various bubble-supported morphologies

• Liftoff mats form a succession of morphologies with bubble deformation of flat mats into tent, ridge, finger, sheet, and strip liftoff mat

• Gas bubbles are a driving mechanism of mat morphology and should be considered when interpreting preserved biosedimentary structures

Supporting Information:

Supporting Information may be found in the online version of this article.

Bubble nucleation modifies the morphology of microbial mats by providing buoyancy, which pulls upward on the mats. If this force is large enough, mats stretch, and deform in response to bubble lift (Bosak et al., 2009;Mata et al., 2012). When enough bubbles accumulate, the mats can lift off the sediment surface and deform into various bubble-supported morphologies (Bosak et al., 2009, Wharton et al., 1983). As the buoyant forces exceed the tensile strength of the mats, the mats rip, exposing the sediment floor. The microbial communities continue to grow with mat morphology modified by the history of bubble lift and deformation. When fossilized, microbial mats with bubble-induced morphologies can be used as paleoenvironmental and paleoecological indicators.

To identify gas-saturated paleoenvironments, we need to document the processes that support bubble production and the resulting morphologies in modern environments. In the biosedimentary record, the interaction between microbial mats and gasses is predominantly described as forming small-scale primary porosity (Broughton, 2024;Donaldson, 1976;Fouke et al., 2000;Gerdes et al., 2000;Guido & Campbell, 2011;Hamilton et al., 2017;Hinman & Lindstrom, 1996;Lynne et al., 2008;Mata et al., 2012;Noffke et al., 2002;Wilmeth et al., 2022). Only a few studies have linked larger scale morphologies to bubble-influence (e.g., Golubic, 1973), and the paucity of studies may indicate that environmental conditions only rarely allowed bubble liftoff of ancient microbial mat. Alternatively, the morphologies that arise from more extensive bubble deformation may have been missed because they have not yet been fully described from modern environments.

Gas-influenced mat morphologies occur in numerous modern environments, demonstrating a breadth of bubble origins and influences on microbial communities. For example, tidal flats and shallow hypersaline basins have cm-to dm-scale microbial mat domes and folds with inverted "U"-shaped tops sometimes referred to as petees (Aref et al., 2014;Gerdes et al., 1993Gerdes et al., , 2000)). These evaporite encrusted petees are interpreted to form by the upward migration of gas from the decomposition of deeper buried organic matter. In supratidal ponds, mats form hollow cones up to 20 cm tall with 10 cm diameter bases (Aref et al., 2014). These cones form through the accumulation of photosynthetic gases and are stabilized by gypsum precipitation (Aref et al., 2014). In the Middle Island Sinkhole in Lake Huron, mats mainly composed of purple filamentous cyanobacteria grow in hypoxic waters and trap bubbles containing methane and sulfide (Biddanda et al., 2015;Nold et al., 2013;Voorhies et al., 2012). The bubbles pull the mat upwards into finger morphologies (Voorhies et al., 2012). Similarly, in laboratory cyanobacterial mats, bubbles pull the mat upwards and remained stable for several weeks (Bosak et al., 2009). Some of the most diverse liftoff microbial mats have been described in the perennially ice-covered lakes in the McMurdo Dry Valleys (MVD). Benthic mats growing in the gas-saturated zones of lakes Fryxell, Hoare, Bonney, and Joyce often contain bubbles that extensively deform the mats (Andersen et al., 1998;Craig et al., 1992;Mackey et al., 2015;Wharton, 1994;Wharton et al., 1983). Some mat completely detaches from the lake floor and floats to the ice-water interface where it can become incorporated into ice cover and mobilized out into the surrounding environment through ablation of the overlying ice (Parker et al., 1981(Parker et al., , 1982;;Wharton et al., 1983). This process results in the export of organic carbon (Parker et al., 1981(Parker et al., , 1982;;Wharton et al., 1983) from closedbasin lakes onto the MDV soils. If the liftoff mats remain on the lake floor and environmental conditions promote early lithification, bubble-induced morphologies can become an initiation point for subsequent growth of lithified stromatolites (Mackey et al., 2015).

In this contribution, we characterize the morphology and distribution of bubble-supported benthic mats in MDV lakes Fryxell, Joyce, and Hoare. Our work spans observations from multiple field seasons and shows that the morphology of liftoff mats is complex, varies with depth and amongst the lakes, and is best explained by protracted bubble-driven deformation. The depth zones supporting liftoff mats have expanded and contracted over time with corresponding shifts in mat morphology. When preserved, resulting composite morphologies can be used to track changes in the lake level, gas supersaturation, and biomass export from the MDV lakes. The liftoff mats described in this study include numerous examples of bubble-induced morphologies that serve as a reference for identifying similar mats in the geologic record and interpret their paleoenvironmental context.

Background

The McMurdo Dry Valleys (MDV) region of Southern Victoria Land, Antarctica (Figure 1), is one of Earth's coldest and driest polar deserts, receiving a maximum of 50 mm water equivalent precipitation per year (Fountain et al., 2010). However, glaciers flow into the valleys, and in austral summer, glacial melt water creates streams that flow to closed basin lakes, including lakes Fryxell, Hoare, and Joyce (Craig et al., 1992;Hawes et al., 2016, Figure 1b). These lakes are perennially ice-covered, and water input is approximately balanced by ablation from the ice surface (Craig et al., 1992;Dugan et al., 2013;McKay et al., 1985), which occurs at an average rate of 30 cm yr -1 (Craig et al., 1992). The streams carry mud-sized sediment into the lakes that gets deposited on the benthic mats (Mackey et al., 2017;Rivera-Hernandez et al., 2019) and mud layers often define annual mat growth laminae (Sutherland & Hawes, 2009).

Open water moats often develop along the lake's margins during the summer months due to absorption of solar energy (Stone et al., 2024). This region is wind mixed and allows atmospheric exchange across the air-water interface (Wharton et al., 1987). However, for most of the lake, the perennial ice cover and salinity stratification maintain stable water columns and inhibit mixing (Hall et al., 2017;Obryk et al., 2016Obryk et al., , 2019)).

Physical and biological processes influence dissolved gas concentrations. Dissolved gases enter the lake through meltwater streams during the summer (Wharton et al., 1987). Atmospheric gases (such as N 2 , O 2 , and Ar) are excluded from the ice as the water freezes (Andersen et al., 1998;Craig et al., 1992) and accumulate in the water, creating gas supersaturation below the ice. These bubbles can be trapped in the ice, and bubble capture of water freezing to ice can account for approximately 70% of shallow water gas loss (Craig et al., 1992). Biological processes produce and remove O 2 due to photosynthesis and respiration. The ice cover transmits 0.5%-3% of the incident light, which supports benthic and planktonic photosynthesis to depths of up to 10s of meters in the lakes (Hawes et al., 2016;Howard-Williams et al., 1998;Jungblut et al., 2016;Sumner et al., 2015) with light compensation irradiance for microbial mats of 0.1% summer surface incident or less (Hawes & Schwarz, 1999;Vopel & Hawes, 2006).

Benthic microbial mats cover the floors of these lakes within the photic zone. The microbial community composition and morphology vary with depth, photosynthetically active radiation (PAR), and water chemistry (e.g., Jungblut et al., 2016). These mats are dominated by filamentous cyanobacteria with other bacteria, some diatoms, fungi, and metazoa (Dillon et al., 2020a(Dillon et al., , 2020b;;Jungblut et al., 2016;Wharton, 1994;Zhang et al., 2015). Morphologies include liftoff, prostrate, and pinnacle mat (Jungblut et al., 2016;Sumner et al., 2015Sumner et al., , 2016;;Zhang et al., 2015). Prostrate mats have mostly flat upper surfaces. In contrast, pinnacle mats consist of mm-scale tufts of filaments to cm-scale peaks separated by prostrate mat (Mackey et al., 2015;Sumner et al., 2015Sumner et al., , 2016;;Vopel & Hawes, 2006). The delicate morphology of pinnacles requires that they grow in stable low-energy environments for multiple years (Mackey et al., 2015;Sumner et al., 2016;Vopel & Hawes, 2006); therefore, this mat morphology is usually found in deeper water, below the depths where mat is disrupted by liftoff processes.

Gas bubbles nucleate in and on microbial mat at depths where dissolved gas concentrations are sufficient (Wharton et al., 1983). In the early 1980s, the zones of liftoff mat formation in lakes Bonney, Fryxell, and Hoare extended from the base of the ice to maximum depths of 8-11 m depth, and float mats (mat that completely detaches from the floor and floats up to the ice-water interface) was documented above the same depths (Parker et al., 1981;Wharton, 1994;Wharton et al., 1983). Bubbles were interpreted to have formed from photosynthetically produced oxygen entrapped by the microbial mats (Parker et al., 1981;Wharton et al., 1983).

Study Sites

Lake Fryxell (77°37′S, 163°11′E) abuts the east side of Canada Glacier and has a maximum depth of ∼20 m (Spigel & Priscu, 1998) with a 2.3-6 m thick perennial ice cover (Priscu, 2023). Water sources include 13 glacial meltwater streams as well as directly from the Canada Glacier (McKnight et al., 1998). Historical changes in lake level resulted in evaporation and refilling events, which produced salinity stratification in the lake (Green et al., 1988;Lawrence & Hendy, 1985;Lyons et al., 2005). Here, all depths for Lake Fryxell reported have been adjusted to 2023 lake levels. The lake is supersaturated with dissolved oxygen at shallow depths and turns euxinic below ∼9.8 m (Jungblut et al., 2016;Lumian et al., 2021). Lake Hoare (77°38′S, 162°52′E) abuts the west side of Canada Glacier. The lake has a maximum depth of ∼36 m (Spigel & Priscu, 1998) and a 2.3-6.0 m thick perennial ice cover (Priscu, 2023). Water sources include the Canada Glacier, Andersen Creek, and drainage from Lake Chad (Hawes et al., 2016;Wharton et al., 1992). Based on water chemistry, Lake Hoare is believed to have drained or evaporated to dryness at c. 1,200 yr. BP (Lyons et al., 1998). Over the past decades, the lake level has been rising (Chinn & Maze, 1982;Doran et al., 2008) except for a period of decreased levels between 1986 and 2002 (Doran et al., 2002). The upper water column is mixed, becomes density stratified from between 12 and 16 m depth, and is weakly stratified below 16 m (Spigel & Priscu, 1998;Sutherland & Hawes, 2009). Temperature, dissolved oxygen, and pH are higher in the upper section (Sutherland & Hawes, 2009). The lake is anoxic below 25-26 m depth (Vopel & Hawes, 2006). Lake Joyce (77°43′S 161°38′E) is ∼55 m deep (Mackey et al., 2018;Spigel & Priscu, 1998) and receives water from glacier streams and Taylor Glacier (Greene et al., 1988). The lake has a freshwater layer below the ice cover to approximately 10 m depth and a stepwise increase in conductivity below 10 m (Spigel & Priscu, 1998). Lake Joyce's level has been rising since the late 1940s at an average rate of 0.2 m y -1 (Mackey et al., 2018). This influx of meltwater has been accommodated by thickening of the shallow freshwater layer (Hawes et al., 2011;Shacat et al., 2004). Dissolved oxygen starts to decline at 21 m and the lake is anoxic below 25.5 m depth (Shacat et al., 2004).

Methods

Documentation of Microbial Mats

Diver observations, video footage, and photographs were used to catalog the microbial mat morphologies and their spatial distribution in each studied lake. To obtain the data, holes were melted or drilled through the ice cover of lakes Fryxell, Hoare, and Joyce, which allowed tethered divers with cameras, a Chasing M2 underwater drone with video and timelapse capabilities, drop cameras, and GoPro cameras in timelapse mode to investigate benthic mats. The dates, modes, and depths of documentation at each location are summarized in Table 1 and expanded on here. Microbial mats were imaged from Lakes Fryxell in December 2006, November 2012, and December-January 2022-2023. In 2006, video transects of Lake Fryxell were collected from 4.2 to 5.0 m depth and a photo transect was collected at 4.6, 5.5, and ∼6.4 m depth. Video transects were collected from 8.5 to 10.5 m depth in November 2012. In December 2022 and January 2023, video transects were collected from 3.0 to 11.0 m depth in and GoPro cameras in 30 s time-lapse mode were placed on tripods on the lake floor to document the movement of mats at 4.3, 6.1, and 7.9 m depths. A photo transect of Lake Hoare was collected from 6. 1, 7.3, 7.9, 8.8, 10.1, and 11.3 m in December 2010. Microbial mats were imaged from Lake Joyce in November 2010 and November 2014. A video transect of Lake Joyce was collected from approximately 5-21.6 m depth in November 2010 from the southwest part of the lake. In November 2014, a photograph transect was collected from 7.0 to 22.1 m in the northern part of Lake Joyce and downward-facing drop and oblique GoPro cameras were lowered through holes drilled in the ice to image the lake floor at 55 sites across the lake with depths ranging from 9.1 to 55.4 m (Mackey et al., 2018).

An example of finger liftoff mat was collected in November 2014 from Lake Joyce. It was frozen and imaged to nondestructively reveal the interior three-dimensional structure. Imaging consisted of X-ray micro-computed tomography (CT) with a Siemens Preclinical Solutions Inveon CT scanner at the UC David Center for Molecular and Genomic Imaging with 196 × 196 × 196 μm voxels. Results were viewed with Dragonfly software.

Nonliftoff and liftoff mat samples were collected from the floor of Lake Fryxell in January 2023 at 4.3 m (n = 4), 6.1 m (n = 4), and 7.9 m (n = 4) depths. Float mats were collected from the ice-water interface above 4.3 m (n = 4), 6.1 m (n = 4), and 7.9 m (n = 5) lake floor depths. The mats were dissected along vertical cross sections in the field to measure mat thickness and number of laminae previously shown to be annual growth markers (Hawes et al., 2016).

Lake Level and Ice Cover Thickness

Lake level (Doran & Gooseff, 2023) and ice cover thicknesses (Priscu, 2023) for lakes Fryxell and Hoare were measured during each austral summer season by the McMurdo Dry Valleys Long Term Ecological Research group from 1991 to present and 1989 to present, respectively. Lake level was measured manually to liquid water level in drilled holes or by surveying a moat connected to the lake water under the ice cover (Doran & Gooseff, 2023). All ice thickness data from the MCM LTER are reported as the distance between the bottom of the ice to the piezometric water level not the irregular top of the ice cover. We used these data to track lake level and ice cover thickness changes over time for lakes Fryxell and Hoare. These data sets were accessed through the MCM LTER website (www.mcmlter.org). Although data from Lake Joyce is more sparse, observations of recent lake level rise, combined with reconstructed bathymetry, provide a 68-year record of lake level change (Mackey et al., 2018).

Results

Changes in Lake Level and Ice Cover

Lake level and ice-cover thickness have fluctuated during the time interval of liftoff observations (Doran et al., 2002;Hawes et al., 2011;Priscu, 2023; this study, Figure 2). Lake Fryxell's level was recorded at 16 m.a.s.l.

(meters above sea level) in 1980-1981, 17.4 m.a.s.l. in 2006, and 18.2 m.a.s.l. in 2022(Doran et al., 2002)). The ice cover was 5 m thick in 1980-1981, the piezometric ice thickness ranged from 5.0 to 5.8 m thick in 2006, and 2.3-2.8 m thick in 2022-2023 (Priscu, 2023;this study). The lake level and ice cover thickness for Lake Hoare have also varied over time. Lake Hoare level was at ∼83.8 m.a.s.l. in 1978-1979(Chinn, 1993)), 73.9 m.a.s.l. in 1996, and 74.5 m.a.s.l. in 2010 (Doran et al., 2002). The ice cover thickness was 5.0-5.5 m in 1978-1979 (Parker (Hawes & Schwarz, 1999), and 2.7-3.4 m thick in 2010 (Priscu, 2023). Lake Joyce's level has steadily increased since at least 1947 (Hawes et al., 2011;Mackey et al., 2018). The ice cover was approximately 5 m thick at the southern site in 2010 (Hawes et al., 2011) and 4.3 ± 0.2 m thick at the northern site in 2014 (Mackey, 2016).

Microbial Mat Morphology and Distributions

We categorized bubble-deformed mats in all the lakes into five liftoff morphologies: tent, finger, ridge, sheet, and strip liftoff mats. Sheet and strip liftoff mats were sometimes observed to completely detach from the lake floor and float upwards (Movie S1). This mat type was categorized as float mat (Parker et al., 1981;Wharton et al., 1983).

Liftoff Mats

Tent liftoff mats consist of microbial mat that is raised vertically at one point, creating a three-dimensional, hollow, tent-like structure, 2-70 cm tall (Figures 3a-3c). Bubbles are largest at the peak, and additional smaller bubbles may be present along the sides of the mat. The tops of tents are rounded and have smaller diameters than the bases, which can be 10s of centimeters. The tent structures are equivalent to lithified columnar liftoff mats described by Wharton (1994).

Finger liftoff mats consist of mat that has a single peak, but unlike tent structures, the diameter of fingers is small (2-3 cm) and stays relatively constant from the base to the top (Figures 3d-3f and 8e; Mackey et al., 2015). Some fingers have a bulbous top surrounding a bubble. Finger liftoffs are up to 20 cm tall. Most fingers occur as single structures, but in some cases, they cluster together or are found extending upward from other liftoff mat morphologies (Figure 3f). Ridge liftoff mats consist of mat vertically raised into a ridge over several centimeters (Figures 3g and 3h). Ridges can be up to 30 cm tall and 50 cm long. The tops of the ridges are rounded but can either be narrow (Figure 3g) or wide (Figure 3i). These liftoff structures have bubbles distributed throughout. These have not been previously described in MDV lakes to our knowledge.

Tent, finger, and ridge liftoff mats are sometimes associated with rips in the mat, creating partial openings between the water and the substratum below the lifted mat. When these tears are at liftoff structure bases, they create open cones in the case of tents (Figures 3a-3c) or flaps in the case of ridges. Tears at the apex leave irregular openings (Figure 3i). Where tears are large enough, the mat forms sheets, consisting of mat suspended in the water column that are up to 50 cm wide and 60 cm tall (Figures 3j and 3k). One side of sheet liftoffs has photosynthetic pigments (Hawes & Schwarz, 1999) similar to the color of nearby in situ mats, whereas the other is tan and lacks pigments. These surfaces are continuous with horizons within the mat subsurface. Strip liftoff mat is a long and thin version of sheet liftoff (Figure 3j). Both often hover above the lake floor with positive buoyancy from bubbles.

Float Mat

Float mats are liftoff mats that detach from the lake surface and float within the water column or at the ice-water interface (Wharton et al., 1982(Wharton et al., , 1983)). Float mats range in size from cm to dm in maximum dimension (Figures 4a and 4b). During the 2022-2023 field season at Lake Fryxell, we documented liftoff mats detaching from the lake floor at depths of 4.3, 6.1, and 7.9 m. These liftoff mats abruptly detached from the lake floor and floated up vertically to the bottom of the ice in less than 2 min (Figures 5a-5f, Movie S1). Float mats moved around under the ice, and some lost enough buoyancy to sink back down (Figures 5g-5j). The sinking mats commonly descended diagonally, landing on the lake floor at a different location than they detached.

Folded and Torn Mat

Folded and torn mats were commonly present on the lake floors at all liftoff sites (Figures 3k, 4d, 4e, and 8c). Smaller folds were present as multiple parallel upright and overturned folds with less than 5 cm relief (Figures 4e and 8c). Larger folds were present as single vertical (Figures 3j and 3k) or overturned (Figure 4d) folds often at the tops of sheet or strip liftoff mats. Some had vertical relief greater than 5 cm. Laterally, folded mats ended abruptly exposing the lake floor.

Mat Thickness and Laminae

The thicknesses and numbers of laminae of all mat types had similar ranges within each depth but varied with depth (Figure 6; Juarez Rivera et al., 2024). Liftoff, float, and in situ mats at 4.3 m depth had thicknesses that ranged from 0.4 to 1.3 cm, and the number of laminae ranged from 2 to 5. At 6.1 m, mat thickness ranged from 0.5 to 3.0 cm, and the number of laminae ranged from 1 to 12. At 7.9 m, mat thickness ranged from 1.3 to 5.0 cm, and the number of laminae ranged from 2 to 11.

Microbial Mat Liftoff Distributions in Lakes Fryxell, Hoare, and Joyce

Liftoff mats extend to different maximum depths, which we have termed the liftoff zone, in the shallow regions of lakes Fryxell, Hoare, and Joyce. Within the liftoff zone, gas bubbles were present at the surface and within the laminae on all liftoff mats, and in some cases, on mats without liftoff. In the latter case, trapped bubbles sometimes deformed the mats with cm-scale blisters but no liftoff was observed (Figure 4c).

Lake Fryxell

The first documentation of liftoff mats at Lake Fryxell is from 1980 to 1981, liftoff mats were present from 5 to 8 m depth in Wharton et al. (1983). In December 2006, the liftoff mats were present from 4.2 to 6.4 m depth and vertical relief increased with depth. At the shallowest regions of the liftoff zone, liftoff morphology was dominated by tent structures with maximum vertical relief of 10 cm (Figure 7a). Many of the tent liftoffs were partially torn. Finger liftoff was present from 4.6 to 4.8 m depth with maximum relief of 13 cm at 4.5 m depth. Ridge liftoff was present from 4.2 to 6.4 m depth. Sheet liftoff was attached to flat mat and was dominant from 6.2 to 6.4 m depth with vertical relief up to ∼35 cm (Figure 7b).

In November 2012, the benthic mats were only documented from 8.8 to 10.6 m depth. Since the shallower depths were not observed, active liftoff mats were not documented. Isolated fragments of pinnacle mat were present at 10.1 m below the zone of pinnacle mat growth (Figure 4f; Sumner et al., 2015).

In December 2022 and January 2023, liftoff mats were documented from 3.5 to 8.5 m depth. Journal of Geophysical Research: Biogeosciences with depth. For example, sheets were up to 27 cm tall and 31 cm wide at 3.5 m depth (Figure 7c), 80 cm tall and 27 cm wide at 6.1 m depth, and 84 cm tall and 91 cm wide at 7.5 m depth (Figure 7e). Tents were present throughout the liftoff zone usually less than 20 cm tall and 30 cm wide. A distinctly large tent, 62 cm tall and 130 cm wide, was present at 6.1 m depth (Figure 3c). Ridges and fingers were the least common of all liftoff morphologies and were present from 3.5 to 6.1 m depth and 4.1 to 5.0 depth, respectively. At depths from ∼7.2 to 8.5 m, pinnacle mat was lifted off to form sheets with fewer tents (Figure 7e), and the float mats above these depths were also composed of pinnacle mat (Figure 4b).

In Lake Fryxell, the liftoff zone was 0.8 m thicker in 1980-1981 (Wharton et al., 1983) than in 2006. By 2022-2023, the liftoff zone was 2.8 m thicker extending significantly into the deeper area of pinnacle growth.

Lake Hoare

In 1978-1979, the liftoff zone was documented from 5.5 to ∼11 m depth (Parker et al., 1981) or 5-∼12 m depth (Wharton et al., 1983). In 1996, the liftoff zone was present from 3.5 to 5 m depth (Hawes & Schwarz, 1999). Thus, the liftoff zone was at least 4 m thicker in 1978-1979 than in 1996, eliminating the potential for bubble disruption for a wide swath of mats. In December 2010, the liftoff mat at Lake Hoare was only observed at 6.1 m depth (Figure 8). The mats were not imaged shallower than this; consequently, the thickness of the liftoff zone at this time is unknown. The mats imaged at 7.3 m did not show evidence for liftoff, so the base of the liftoff zone was intermediate between that observed in 1978-1979 and 1999. At 6.1 m, the lake floor was mainly covered by partially torn tent and finger liftoffs as well as sheet liftoffs (Figures 8a and 8b). Ridge liftoff was less common.

Tents had maximum vertical relief of 15 cm, and the base diameter reached up to 55 cm wide in partially torn tents. Finger structures had vertical relief of less than 10 cm. Fingers were sometimes superimposed on partially torn mats (Figure 8b). Tear scars were also present (Figure 8a). Visible bubbles became less common with increasing depth. Fewer bubbles were present on microbial mats at 7.3 m where they caused deformation without liftoff (Figure 8c). At 8.8 m depth and below, the mats transitioned to pinnacle morphologies without liftoff or bubbles (Figure 8d).

Lake Joyce

Microbial mat liftoff at Lake Joyce was documented in the southwest part of the lake in November 2010 (Figures 9a-9c) and in the northern part in November 2014 (Figures 9d-9k). The liftoff distribution and morphologies varied between these two locations at the two different times.

In November 2010, liftoff mats were observed between 6.9 and 7.9 m depth, they were not present between 6.9 and ∼5 m, which was at the base of the thick ice cover (Mackey, 2016). In the liftoff zone, thicker, pink-colored mats and thinner, purple-colored mats covered most of the lake floor, and liftoff mats were composed of both mat types (Figures 9a-9c). Tent and ridge liftoffs were more common with maximum vertical relief of ∼25 and 10 cm, respectively. Partially torn tents were present at the top of the liftoff zone. Finger liftoff was not observed at this location. Folded mat and tear scars were present throughout the liftoff zone (Figure 9c). Liftoff mats laterally coexisted with pinnacle mats, and in some cases, tent liftoffs had pinnacles on their surfaces with pinnacles oriented normal to the tent surface (Figure 9b).

The northern location, observed in November 2014, was near a delta created by meltwater stream J3 (Green et al., 1988;Mackey et al., 2015). Here, active liftoff mats were only observed at 7.0 m depth below the 4.3 m thick ice cover (Hawes et al., 2011). Collapsed mats were present at 9.4 and 10.6 m depth. At 7 m depth, all liftoff mats comprised thin purple-colored mats (Figure 9d). The mats sparsely covered the lake floor compared to the 2010 location as well as to lakes Fryxell and Hoare. Liftoff mainly occurred as fingers, tents, partially torn tents and fingers with some sheets. Ridges were present but rare. Nonliftoff mat was laterally continuous with liftoff mat. Maximum vertical relief of finger liftoff was 21 cm, of tents was 6 cm, and of sheets with fingers was 64 cm.

Figure 4.

Representative images of microbial mat textures related to liftoff mat. Liftoff mats can detach from the lake floor and float to the ice-water interface (a, b). Float mat was composed of pinnacle mat between 7.5 and 8.5 m depth at Lake Fryxell (b). Gas bubbles modify the texture of microbial mats without causing liftoff (c). Mats collapsed and folded with the loss of buoyancy (d, e). The transport of liftoff mat left tear scars on the portion of mat remaining on the lake floor. When liftoff mat is mobilized the remnant mat exhibits tear scars and folding (e). Pinnacle mat at 10.3 m depth in Lake Fryxell below the growth zone for pinnacle mats (f). This section of pinnacle mat is folded and was located deeper than pinnacle mat grew. Thus, it likely fell into deep water after losing buoyancy (Figure 5). Example of liftoff mats exported through the ice cover (g-i).

Journal of Geophysical

Research: Biogeosciences 10.1029/2024JG008516 JUAREZ RIVERA ET AL. A CT scan of a finger liftoff revealed air pockets at the bulbous top and along the sides with sediment concentrated on one side of the finger (Figures 9e-9g). These mats had fewer bubbles visible on the surface than Lakes Fryxell and Hoare, and the mat texture was smoother than in Lake Hoare and the southern location of Lake Joyce. Some larger bubbles were present on top of the beige substratum left behind by earlier delamination and appeared to be stabilized by thin biofilms. Collapsed tents, ridges, and folded mats with pinnacles perpendicular to these surfaces were present at 9.4 m (Figures 9h and 9i). Ridges and potential collapsed tents with upwardly oriented pinnacles were present at 10.6 m depth (Figures 9j and 9k).

Discussion

The variety of liftoff morphologies and their changing distribution through time allow us to connect the morphological effects of bubble disruption to microbial and environmental processes in the lakes.

Model of Liftoff Processes

The morphology of liftoff mats in lakes Fryxell, Hoare, and Joyce is interpreted as highly influenced by the history of the nucleation of bubbles on initially prostrate or pinnacle microbial mats. In our model, the liftoff process starts with the buoyant lifting of mats, which causes a vertically oriented tension. When this tension exceeds the cohesive strength of the mats, they delaminate along a weak layer, usually a sediment-rich layer or the base of the photosynthetically active zone. Tent and ridge liftoff mats form as primary liftoff morphologies when delamination occurs 3g,and 3h). Because the surface areas of tents and ridges are larger than they were prior to liftoff, we infer that the mat also stretches in response to shear stress induced by bubble buoyancy. In some cases, shear continues, the mat continues to extend, and the peaks get higher, forming finger liftoff morphologies (Figures 3d-3f). However, in most cases, the mat rips because the buoyant forces exceed its tensile strength. Ripping leads to the formation of open cones and ridges (Figures 3a,3b,and 3i). Ripping can also shift the bubbles and sometimes release them, changing the distribution of buoyancy stresses. Loss of bubbles can cause collapse of the structures into folded mat on the lake floor (Figures 4d and 4e), whereas shifting of bubbles changes liftoff morphology, sometimes causing additional tearing. When one side of a tent or ridge completely rips off the bottom, sheets form (Figures 3f,3j,and 3k). This morphology is open enough that any bubbles under the mat are likely released. Thus, we predict that many sheets lose at least some of their buoyancy when they first form (Figure 3k). If they regain buoyancy with expansion or growth of bubbles, they may rip in new places, forming narrower strips of liftoff mat (Figure 3j). If the liftoff forces become strong enough relative to the remaining mat attachment points, the mats completely detach from the lake floor and become float mats (Figures 4a, 4b, and 5a-5f, and Movie S1). Some float mats remain at the ice-water interface when new ice starts to form and freeze into the ice cover to be later exported from the lake. Others lose their bubbles and buoyancy, sink to the lake floor (Figures 5g-5m), and form a substratum for regrowth of the mat. Mat growth also continues at all stages of liftoff formation, modifying mat topography and ripped mat edges. Thus, liftoff structures often become scaffolding for future microbial mat growth. The combination of all these processes produces diverse mat morphologies that reflect bubble-induced deformation and mat responses. The detailed distribution of morphologies that result from bubble buoyancy reflects both microbial and environmental processes.

A Model for Liftoff Mat Morphology

Comparative analyses of liftoff mats in lakes Fryxell, Hoare, and Joyce over time suggest a model for liftoff mat morphology development in which gas bubbles variably deform benthic mats into large-scale features (Figure 10). At the undeformed end of this spectrum, mat morphologies are not affected by gas bubbles, for example, mats that grow at depths where dissolved gas concentrations are not sufficient to form and maintain bubbles. In environments, where small bubbles nucleate within similar mats, they exhibit slight modifications by bubbles such as the development of mm-to cm-scale high surface irregularities (Figure 4c). In these mats, the subcentimeter texture shows ample evidence of gas-modification, which can be preserved as fenestrae (e.g., Mata et al., 2012), which is an important distinction between mats free from bubble influences and those with the influence.

Where bubbles provide buoyancy that exceeds the mat tensile strength, macromorphologies develop that are the direct result of mat modification by gas bubbles such as tent and ridge liftoff. These morphologies started with benthic mats attached to the lake floor that delaminated. The details of their morphological development depend on the details of buoyancy versus mat strength. Many mats stretch upward to develop steep walls and hollow interiors but remain attached to the sediment surface. Some of these mats delaminate over a wide area, creating a wide base. However, others stretch and extend upward over a narrow base to form fingers. The mat characteristics that preferentially support the formation of tents and ridges in contrast to fingers may be related to the thickness of the mat. Specifically, finger liftoff was common at Lake Joyce in thin purple mats (Figure 9d) that may have been easy to shear along the laminate. In contrast, we did not observe finger liftoff with pinnacle mat, which tends to be thicker than prostrate mat.

When the shear forces exceed the tensile strength of the mats, they begin to tear, which modifies liftoff morphologies. Tearing reduces the attachment of tent, finger, and ridge morphologies to the lake floor, and they can convert into sheets. The size of sheet mats often exceeds the sizes of the previous features because the buoyant forces get transferred to adjacent nonliftoff mat attached to them, and these can delaminate and also be pulled upward. At the end of this deformation trajectory are strips of mat. Strip liftoff likely forms from multiple cycles of deflation and buoyance with changing bubble conditions, which causes additional tearing of the mat. The effects of these cycles of loss and regrowth of bubbles were observed at Lake Fryxell in 2022-2023. Abundant sheet and strip liftoff was present, especially from 3.5 to 7.5 m depth when and where bubble formation was most abundant. In contrast, 2006 liftoff mats in Lake Fryxell only had sheet liftoff at 6.2-6.4 m depths. The greater abundance of sheet liftoff in 2022-2023 suggests that the dissolved gas concentration was greater, creating more bubbles and greater stress in the mat. The large concentration of strip liftoff mat suggests multiple cycles of tear and liftoff as bubbles were released and regrew.

Although the bubble forces are essential for producing liftoff morphologies, the microbial communities are responsible for cycling gases and the tensile strength of the mats. The shallow water benthic mats in our study contain abundant filamentous cyanobacteria classified as Oscillatoriales including Phormidium, Leptolyngbya, with Pseudanabaena in lower abundances (Jungblut et al., 2016;Zhang et al., 2015). Cyanobacterial community composition varies among and within lakes, and some morphological variations have been attributed to different microbial community compositions (Harwood & Sumner, 2012;Hawes et al., 2011;Mackey et al., 2015;Sumner, 1997). However, all of these communities produce oxygen through photosynthesis, which contributes to the formation of bubbles by increasing the supersaturation of gases in the water. The rate of oxygen production depends on environmental conditions such as the available irradiance and nutrient availability. Thus, only specific environmental conditions allow enough photosynthesis for microbial processes to increase gas saturation. Thus, bubble formation and mat liftoff reflect both microbial processes and environmental conditions.

Liftoff Mats as Environmental Change Indicators

The extent of liftoff mat development and deformation depends on the stability of an active liftoff zone. Our results show that the liftoff zones have fluctuated through time with the potential to change the extent of bubble lift of microbial mats at any point in this model.

When the liftoff zone shifts across depths with different mat textures, the mats develop distinct morphologies that can be used to track both the extent and directional change of liftoff zones through time. Pinnacle orientation can be used to identify disrupted mats because pinnacles often grow vertically. When they grow on horizontal surfaces, they grow at 90°to the surface. However, when they grow off tilted surfaces, they often still grow close to vertically (Mackey et al., 2017). Thus, pinnacles will be tilted by the same amount as the surface if they grew before the deformation, and if they grew after the deformation, they will be oriented more vertically. Because it takes multiple years to develop the pinnacle morphology as demonstrated by counting annual laminations (Mackey et al., 2015;Sumner et al., 2016;Vopel & Hawes, 2006), old liftoff mats can be distinguished from recent liftoff pinnacle mats. In lakes Fryxell (Figure 7e) and Joyce (Figure 9b), pinnacles were oriented perpendicular to the surface of the tent liftoff mats. Hawes et al. (2011) also documented contorted liftoff mats with pinnacles oriented perpendicular to the folds at 6 m depth in Lake Joyce. The liftoff of pinnacle mats in these cases is best explained by multiyear growth of pinnacles followed by an expansion of the depth of gas superpsaturation and bubble nucleation into the zone of pinnacle growth. This intrusion facilitated the formation of gas bubbles in the pinnacles, which subsequently raised the pinnacle mats into tent liftoff mats with tilted pinnacles. The expansion of the liftoff zone occurred despite the overall increasing water depth at both lakes (Hawes et al., 2011). In both lakes, the formation of liftoff structures in pinnacle mat requires an increase in bubble formation at depths where pinnacle mat grew without disruption for years. This could be caused by either lower hydrostatic pressure or increases in gas concentrations. Histories of lake level show that it did not decline in these years, so higher gas concentrations are required. This could have been caused by increased photosynthesis possibly due to thinner ice allowing more irradiance into the environments.

Figure 9. Liftoff mat at Lake Joyce from 2010 (a-c) and 2014 (d-i). Liftoff mats from 2010 are thicker, pink and purple colored and mostly form tent structures (a-c), but ridges and folded mats were also present. Liftoff transitioned to pinnacle mats and some tents had pinnacles on the surface (c). The liftoff mats from 2014 at 7 m depth were thinner and purple-colored mainly forming finger, tent, and sheet liftoff (d). Finger liftoff collected from 7 m depth (e) and vertical CT scan slices (f-g). The CT scans revealed primary porosity at the top and along the sides of the finger with dispersed sediment. Images of collapsed liftoff mats from 9.4 m (h) with zoomed in inset (i) and 10.6 m (j) depth with zoomed in inset (k). At 9.4 m depth, pinnacle mats were present and mostly oriented perpendicular to the surface of liftoff mats. At 10.6 m depth, liftoff structures appeared subdued from mat overgrowth and pinnacles were oriented vertically.

Journal of Geophysical Research: Biogeosciences 10.1029/2024JG008516

When the liftoff zone narrows, any liftoff or textures related to liftoff mats that remain below the new liftoff zone can become scaffolding for future microbial mat growth (Figure 9d). For example, the Lake Joyce liftoff mat documented from 7.6 to 7.9 m depth in 2010 was at a similar absolute elevation as the collapsed mats from 9.4 m depth in 2014, when accounting for the 1.4 m of lake level rise (Priscu, 2023). All these mats contained pinnacles, but the 2010 pinnacles, interpreted as within the liftoff zone, were perpendicular to the liftoff mat surface, whereas by 2014 more of the pinnacles were vertically oriented. The vertical orientation of pinnacles in 2014 suggests that this depth was no longer being disrupted by liftoff, allowing the pinnacles to grow vertically for up to 4 years using collapsed mat as scaffolding. Evidence of even older liftoff is present in lithified stromatolites from Lake Joyce, which have inner smooth tubes with bulbous tops that transition to branches (Mackey et al., 2015). The tubes and their reconstructed growth depth are consistent with finger liftoff, whereas the vertically oriented branches and depth location of the stromatolites are consistent with growth below the liftoff zone (Mackey et al., 2015).

In all cases, the superposition of liftoff mat by vertically oriented branches, pinnacles, or prostrate mat are consistent with microbial mats that began growing within the liftoff zone and continued to grow below it, accreting a new morphology. Thus, we interpret these secondary structures as recording prior liftoff processes from when environmental and ecological conditions allowed high dissolved gas concentrations in the lake water at shallow enough depths for bubble nucleation. The liftoff zone can contract when hydrostatic pressure increases or gas saturation decreases due to lower rates of photosynthesis. Increases in lake level both increases hydrostatic pressure and decreases irradiance. Increased ice thickness also reduces irradiance. Thus, both of these environmental changes would reduce the extent of the liftoff zone.

Preservation Potential of Liftoff Structures

The preservation of liftoff structures like those described in this study requires a combination of long-lasting structures and synsedimentary mineral precipitation. The formation of liftoff structures requires cohesive but not lithified mat, yet lithification is required for their geologic preservation. Lakes Fryxell, Joyce, and Hoare are supersaturated with respect to calcite (Clayton-Greene et al., 1988). However, only a few of the liftoff structures observed were lithified. In some cases, calcite precipitated at the mat-bubble interface on fingers preserving bubble and microbial filament molds (Mackey et al., 2015). In other cases, calcite precipitated on the underside of the mats forming liftoff tents, maintaining hollow interiors even after bubbles redissolved (Parker et al., 1981;Wharton, 1994;Wharton et al., 1982). In both modes of lithification, the preservation of liftoff morphologies and bubble molds required that mineralization happened while the mats were buoyant, and the bubbles were still present. The preservation window for liftoff structures might be extended by microbial processes; the stability of bubbles is increased if enmeshed by microbial biofilms (Bosak et al., 2010). However, even if the bubbles have redissolved or escaped the liftoff structures, evidence of liftoff mat can be preserved. Our observations from lakes Fryxell and Hoare show that even if liftoff mats partially collapse with narrowing of the liftoff zone and loss of bubbles, a version of these morphologies can persist for multiple years, providing sufficient time for subsequent microbial mat growth and longer-term lithification to preserve them in the rock record (e.g., fossilized fingers and Journal of Geophysical Research: Biogeosciences 10.1029/2024JG008516 JUAREZ RIVERA ET AL.

columnar stromatolites Mackey et al., 2017;Wharton, 1994, respectively). By strategically looking for liftoffrelated structures in the rock record, we may discover that their preservation is more frequent than currently documented.

Implications for Bubble-Influenced Microbial Mat Morphology

The morphology of biosedimentary structures has long been used to interpret microbe-environment interactions because larger-scale morphologies have better long-term survival than chemical and microstructural components over geologic timescales (Allwood et al., 2006;Bosak et al., 2013;Riding, 2011). A major challenge in this regard is our ability to link structure to processes and few studies have focused on the role of gas bubbles in shaping the morphology of biosedimentary structures. We predict that there are many more examples of liftoff structures in the rock record given the diversity of modern environments containing bubble liftoff structures (Aref et al., 2014;Bosak et al., 2010;Fouke et al., 2000;Gerdes et al., 2000;Guido & Campbell, 2011;Hamilton et al., 2017;Hinman & Lindstrom, 1996;Lynne et al., 2008;Mata et al., 2012;Noffke et al., 2002;Voorhies et al., 2012). In particular, glacial lakes are an obvious ancient environment where these structures are expected, although lithification of them may be rare. Other lacustrine environments may also reach gas supersaturation, as seen in modern environments, and some include abundant calcite precipitation. For example, 2.7 Ga lacustrine stromatolites preserve bubble molds interpreted as due to oxygenic photosynthesis (Wilmeth et al., 2019(Wilmeth et al., , 2022)). These stromatolites do not include reported liftoff structures, which may be due to low oxygen saturation in the water prior to oxygenation of the atmosphere. However, the absence of a morphological model for liftoff structures may have caused these structures to be overlooked. Even so, lake deposits that accumulated after oxygenation of the atmosphere are more likely to contain evidence of bubble liftoff mats.

Both lacustrine and marine deposits associated with the Paleoproterozoic and Neoproterozoic snowball Earth glaciations are important places to look for bubble-induced deformation with stromatolites in "cap carbonates" a promising target. For example, the mid-Cryogenian Rastof Formation, Namibia, has extensively deformed microbial laminae (e.g., Le Ber et al., 2013, 2015;Pruss et al., 2010). Deformation has been attributed to dewatering (Pruss et al., 2010) or, in areas lacking synsedimentary dikes, deformation has been interpreted as due to storm activity or seismicity (Le Ber et al., 2013). The Tonian Beck Spring Formation, California and Nevada, has some similarly deformed microbial laminae (Harwood & Sumner, 2012) as well as fossilized bubble molds (Mata et al., 2012). We speculate that bubble lift is an alternative model that should be considered for deformation of microbial laminae in both formations.

When we apply our morphological modification model of liftoff morphologies to conical stromatolites, we see some similarities and differences. Liftoff mats with tent morphology are superficially similar to Archean and Proterozoic conical stromatolites (Allwood et al., 2006). Conical stromatolites are laminated structures with semirounded bases that taper toward the top and form by the upward growth of microbial mats in response to chemical or energetic gradients (Allwood et al., 2006(Allwood et al., , 2009;;Bosak et al., 2013;Cloud & Semikhatov, 1969;Petroff et al., 2013;Tice et al., 2011). The hollow interior of tents differentiates them from the continually laminated stromatolites, indicating each structure formed through a different process. However, as our model shows, gas bubbles can overprint previous motifs to create composite structures. This is the case for conical stromatolites with barite-filled bubble cavities at the apex (Donaldson, 1976). In this example, mat layers trapped gases and formed domes parallel to laminae (Golubic, 1973;Noffke et al., 2002). Conical stromatolites usually have laminae that thicken at the apex and the trapped gases would have exaggerated the thickness of individual lamina similar to finger liftoff formation, while maintaining the overall conical shape.

Our findings also suggest an additional process for the formation of folded, roll-up, and torn mats. These morphologies can form as wind-induced mat deformation structures in tidal zones (Bouougri & Porada, 2012;Cuadrado et al., 2015), desiccation-induced features in intertidal zones (Cuadrado et al., 2015), dewatering (Pruss et al., 2010), storm-driven deformations in marine environments (Le Ber et al., 2013;Simonson & Carney, 1999), seismic deformation (Le Ber et al., 2013), or as a response to variations in deep water sediment loading (Schroeder et al., 2009). Our work shows that these features can also form in shallow low-energy environments driven by bubble-induced deformation. This process for creating deformed mats has different environmental implications than the other mechanisms. Specifically, it requires the generation of gas in shallow enough water depths that bubbles can nucleate. Thus, if evidence of bubbles can be found and their composition can be constrained (e.g., Wilmeth et al., 2022), we may be able to glean new insights into microbial metabolisms and water chemistry through time. The preservation of these features requires specific sedimentary environments and processes, such as those found in ice-covered lakes (Rivera-Hernandez et al., 2019). By searching for suites of similar features, we may be able to identify and better understand similar mat deformation and environmental settings in the rock record.

Conclusions

Our study shows that microbial mat liftoff structures in gas-supersaturated regions in ice-covered Antarctic lakes form due to microbial community responses to the stresses provided by bubble buoyancy. Initially, buoyancy stretches the mat as it delaminates along weak horizons. The mat can then continue to stretch to form finger morphologies. However, it more often rips, which can lead to the loss of bubbles. The loss of bubbles, their expansion, and continued microbial growth all lead to morphological variations. When mat rips off the lake floor, it becomes float mat, and when enough bubbles are lost in any liftoff mat, it collapses to the lake floor forming folded mat. These processes produce five buoyant macroscopic mat morphologies that are distinct from deeper water microbial mats not influenced by gas bubbles in the same lakes. These bubble-driven morphologies include tent, ridge, finger, sheet, and strips of mat with associated folded and torn mats. We suggest that these mat morphologies develop as a continuum of bubble-driven deformation starting with prostrate or pinnacle mat forming tent and ridge liftoff structures; then pulling and ripping transforms these into sheet liftoff, while repeated cycles of bubble formation and loss induces additional ripping, which produces strip liftoff. The development of liftoff mats along this continuum depends on the stability of the liftoff zone, which varies over time. Changes in the liftoff zone and lake level can result in the superposition of liftoff and nonliftoff mat morphologies, which can be preserved and be used to track environmental change.

Our improved understanding of liftoff mat morphologies and the processes producing them will allow identification of more gas supersaturated paleoenvironments. Understanding how bubbles deform mats, the environmental conditions that promote liftoff of mats, and the microbial roles in gas cycling can strengthen interpretations of paleoenvironments and paleoecology. We predict that bubble-influenced macromorphologies are preserved in the rock record even though there is a paucity of studies reporting them. Better documentation of these modern analogs may help in their discovery.