On the ability of proglacial lake diatoms to reconstruct Antarctic past ozone changes
Anna Beatriz Jones Oaquim, Heitor Evangelista, Gleyci Aparecida Oliveira Moser, Marcus Vinícius Licínio, Emília Correia, Domênica Teixeira de Lima, Sergei Verkulich, Zinaida Pushina, Yuri Kublitsky
TL;DR
This study shows that diatoms in lake sediments can reveal past ozone changes in Antarctica, indicating recent ozone depletion is unprecedented in 7,700 years.
Contribution
A novel diatom-based proxy for reconstructing Antarctic ozone variability over the Holocene is developed and validated.
Findings
A diatom assemblage shows significant responses to measured ozone data, enabling a quantitative reconstruction model.
Recent ozone depletion is unprecedented compared to the past 7,700 years based on sediment core analysis.
Lake-sediment diatoms are effective proxies for reconstructing stratospheric ozone dynamics in Antarctica.
Abstract
The depletion of Antarctic stratospheric ozone since the 1970s, and the resulting increase in UV radiation reaching the Earth’s surface, have posed a well-recognized threat to polar aquatic and terrestrial ecosystems. Although this phenomenon is primarily driven by anthropogenic emissions, natural processes linked to volcanic activity and changes in solar irradiance can also influence ozone levels over time. Understanding past ozone changes over Antarctica is therefore essential for constraining the amplitude of its natural variability. Given the sensitivity of diatoms to different environmental conditions, we investigated the potential of these organisms as proxies for ozone variability by analyzing their relative abundance along a proglacial lake sediment profile dated using excess 210Pb. We found that a specific diatom assemblage dominated by Gomphonema sp., Nitzschia cf.…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Fig 1
Fig 2
Fig 3
Fig 4
Fig 5
Fig 6- —INCT Criosfera
- —IAEA (International Atomic Energy Agency)/FAO
- —FAPERJ (Rio de Janeiro State Research Support Foundation)
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsPolar Research and Ecology · Geology and Paleoclimatology Research · Marine and coastal ecosystems
Introduction
The stratospheric ozone depletion in Antarctica has been recognized to be one of the most prominent environmental hazards created by man [1]. The unprecedented low concentrations of ozone, first reported in 1985 [2], are attributed to man-made emissions of chlorine and bromine containing compounds, such as chlorofluorocarbons (CFCs) [3,4], methyl bromide (CH_3_Br), hydrochlorofluorocarbons (HCFCs), carbon tetrachloride (CCl_4_) and methylchloroform (C_2_H_3_Cl_3_) [5]. Although this “anthropogenic origin” has gained widespread acclaim and mobilization from the society and the scientific community, natural sources are also players in the photochemical cycle involving ozone destruction. At least two other factors may play a role in the ozone depletion in Antarctica: i. large volcanic eruptions (those of global impact [6] and others occurring within the Antarctic Circle, such as the Mount Erebus [7]); ii. solar activity variability, particularly the 11-year solar cycle, which influences the levels of UV radiation (UVR) reaching the Earth. UVR wavelength bands are classified as UV-A (315–400 nm), UVB (280–315 nm) and UVC (100–280 nm). Among these, UVC—which plays a key role in the photochemical production of ozone in the upper stratosphere— is the most harmful band. However, it is entirely absorbed by the Earth’s atmosphere and does not reach the surface. Of the remaining two bands, UVA undergoes the least atmospheric attenuation and reaches the surface in greater amounts, while UVB—which represents a significant threat to ecological systems [8] and a risk to living biota in terrestrial, oceanic and freshwater ecosystems [9]—is mostly attenuated due to absorption by the stratospheric ozone layer [10].
Impacts on polar environments [11,12] include numerous effects on living organisms [10] at several ecological stages (e.g., growth and reproduction [13], biomass and productivity, community structure [14], photoinhibition [15], and impacts on photosynthetic pigment production [13]). However, these impacts depend on the radiation levels to which organisms are exposed, their specific tolerance (variable among species), their ability to physiologically acclimate and repair possible damage [16], and the vulnerability of the cell. Early studies considering the impact of UVB in Antarctic open waters found that biological effects can be detected to approximately the top 20 m during the spring season, while the most significant injuries were observed within the top 10 m, with DNA being a primary lethal target for UVB [17].This attenuation pattern of incident radiation makes phytoplanktonic organisms from the euphotic zone more sensitive to, and negatively affected by, UV radiation [14] than benthic species. In addition, Antarctic lakes are often very clear (oligotrophic), meaning that the water body itself does not represent an efficient barrier to block the UVB. Among the organisms potentially affected are diatoms (Bacillariophyta), unicellular chlorophyll a/c microalgae, which together make a significant contribution to primary productivity in the Southern Ocean and the global carbon cycle [18].
Diatoms constitute the dominant class of phytoplankton, distinguished by their remarkable species richness [19], and are characterized by a cosmopolitan distribution, encompassing ecosystems from tropical and subtropical regions to polar environments. Their considerable diversity of lifestyles and adaptive strategies allows them to inhabit a wide range of aquatic environments, including marine, freshwater, and brackish systems [20]. Consequently, changes in abiotic or biotic conditions may lead to shifts in diversity and/or abundance of these organisms in the water column [21] and in surface sediments. This sensitivity reflects the responsiveness of diatoms to environmental variation [22]. Several studies have demonstrated that the long-term preservation of diatoms in sediment layers is attributed to their frustules— a cell wall composed of rigid and amorphous silica that are resistant to dissolution [21].These characteristics enable the use of these organisms in paleoenvironmental, paleoclimatic, and paleoecological studies (e.g., Saupe and Mosimann [23]; Katsuki and Takahashi [24]).
The unique morphological and physiological features of these organisms determine how they interact with light. The occurrence of very regular arrays of chambers (areolae) and pores (cribra) that form periodic patterns [25] on the diatoms frustules, in addition to enabling species differentiation, also exhibits certain characteristics of photonic crystals. These periodic and quasi-periodic nanostructures have been shown to turn diatoms frustules into small spectrographs capable of diffracting and refracting UVR while allowing visible light to pass through [26,27]. These nanostructure interact with light, redirecting it toward the cell interior, even when light strikes the frustules at non-normal angles, offering a potential protective mechanism against UVR [27,28]. Other important factors to consider include the refractive contrast between the frustule and the cellular medium, the orientation of the valves, and the degree of the frustules surface being covered by organic matter, for example [27]. Taking into account all off these factors, in the last decades numerous studies have been conducted to understand how those different pores’ and chambers’ arrangement can influence the scattering and reflection of UVR by the frustules from different species. Aguirre et al. [26] for example, analyzed frustules from Navicula perminuta, Coscinodiscus wailesii and Coscinodiscus cf. radiatus in air and observed that scattering and reflection of UVR were relevant, although when the authors used water as the internal medium of the diatoms, simulating natural habitat conditions, they observed that the refractive index was much smaller, and concluded that scattering and reflection effects must be smaller.
Sub-Antarctic lakes, where freshwater diatoms are among the most abundant algal groups [29], provide an unique opportunity to investigate their response to increased UVB radiation under natural conditions as a result of the ozone depletion (Fig 1a) (e.g., Vincent and Roy [30]; Roy [16]). In the Maritime Antarctic, modern and Holocenic proglacial lakes are found, where diatoms are abundant and diverse [31], generally characterized as oligotrophic systems with extremely low water column productivity, except for those influenced by birds and seals that provide nutrient inputs through fecal material [32]. In the benthic lake zone, where efficient regeneration of nutrients can occur [33], productivity and biomass greatly provide better conditions for diatoms development.
Study region (a) Ozone depletion area for august 3, 2023 from Nasa Ozone Watch; (b) King George Island, from Arcgis; (c) Location of lake sediment coring sites at Fildes Peninsula; (d) Geographersee Lake during summer (December, 2024); Profound Lake marked by the red line Google Eart Web during (e) summer without snow (February 20, 2006); and (f) winter with snow and ice (October 19, 2005).
Despite the successful implementation of the Montreal Protocol in 1987, which substantially reduced man-made emissions of chlorine compounds, the Antarctic ozone hole remains in a slow recovery phase and continues to represent a unique global phenomenon. Nevertheless, our understanding of its natural variability prior to the satellite era remains limited. Here, we investigate whether sedimentary diatom communities can serve as indicators of past changes in stratospheric ozone and surface UVB radiation in the Antarctic environment.
We hypothesize that temporal shifts in stratospheric ozone concentrations and the resulting changes in surface UVB radiation, have influenced the composition and structure of diatom assemblages in Antarctic proglacial lakes. Periods of elevated UVB exposure likely favored more UV-tolerant taxa, whereas intervals characterized by higher ozone levels and reduced UVB would have allowed more sensitive species to thrive.
To test this hypothesis, we analyze the relative abundance of diatoms in two dated sediment cores from proglacial lakes on King George Island: one representing the modern period and capturing ozone variability since 1958, which was used for calibration, and another extending back to the middle-to-late Holocene. By assessing whether variations in diatom assemblages track known fluctuations in UVB/ozone in the calibration core, we evaluate the potential of these proxies to reconstruct past UV- levels reaching the surface and, by extension, past stratospheric ozone concentrations. This approach offers a promising avenue for constraining the natural variability of ozone in the pre-industrial Antarctic atmosphere.
Methods
Study area
King George Island (KGI) (61°54’–62°16’S/ 57°35’–59°02’W) (Fig 1b) is the largest island of the South Shetland Archipelago, with a total area of approximately 1,150 km² [34]. It is located 120 km off the northern tip of the Antarctic Peninsula, bounded to the north by the Drake Passage and to the south by the Bransfield Strait [35]. The local climate is classified as cold and moist maritime [36], with mean monthly temperatures ranging from −13.8°C in July and +2.8°C in January. The mean annual air temperature was −2.18°C for the period 1980–2019, as recorded in Bellingshausen Russian Station [37]. The island experiences persistent cloudy conditions with a mean annual precipitation of 684.2 mm for the period 1980–2017 [37]. Cyclonic systems migratie over the Drake Passage and the Antarctic Peninsula with weekly frequency, transporting both polar and subtropical air masses into the region. The largest ice-free area on KGI is the Fildes Peninsula (62°13’S, 58°57’W), which hosts several proglacial lakes (Fig 1c) formed as a result of the millennial-scale retreat of the Collins and Bellingshausen glaciers [38].
Profound lake
Profound Lake (62°11,066’S; 058°54,413’W; Antarctic Place-names Committee – APC) also known as Ozero Glubokoye (Russia), Tiefersee Lake (Germany), Uruguay Lake (Uruguay) and Glubokoye Deep Lake [38,39] (Fig 1e,f), has a surface area of 69.7 m² [40], a maximum depth of 14.7 m [39], and is located 624 m from the Collins glacier [40]. The lake exhibits seasonal stratification of the water column, driven by freezing and thawing processes, with a temporary surface ice/snow cover formed during the winter-spring seasons [38] (maximum thickness of 1 m). Water mixing and oxygenation are relatively limited [41]. The lake formed approximately 4,000 years B.P. following regional deglaciation [42]. Its sedimentary sources are mainly related to wind transport, meltwater from permafrost, and seasonal snowmelt from surrounding slopes [43]. The lake lies on a marine abrasion terrace [44] and is characterized by sediments composed of clay, silt, sand and some submerged mosses [31]. Despite being used as freshwater source for Artigas Base, human impact in the lake is minimal, with no evidence to suggest that it influences the microbial communities [45].
Geographensee lake
The second investigated lake, Geographensee Lake (62°13′00.0ʺ S, 59°01′00.0ʺ W, 40 m 86 a.s.l; Fig 1d), is also located on Fildes Peninsula. The lake has a maximum depth of 4 m and is located 7.21 km southwest of the Collins Glacier [40]. Its catchment area covers 0.2 km^2^ and consists of flat terrain with smooth descending slopes and hills of 65 m in height. Vegetation within the catchment is sparse, occurring primarily as patches of mosses concentrated in moist microtopographic depressions and in the channels of temporary streams that convey meltwater from small snowfields [46].
Local environmental conditions over the lake
Both lakes, which are approximately 7 km apart, experience a seasonal surface snow cover at their surfaces that modulates the interaction of the UVB with the aquatic microbiota. During the end of the austral spring season, both air and soil temperatures reach the melting point, initiating the melting of most of the lake surface. This condition persists until the beginning of fall, when it starts to freeze. Therefore, the period of maximum UVB reaching the Antarctic surface due to annual ozone depletion coincides with the active period of lacustrine life. During winter, higher snowfall and lower air temperatures provide a temporary snow and ice cover, reaching approximately 1 m in depth of ice, as measured in October 2018. In contrast, summer conditions, characterized by lower snowfall and above-zero air and soil temperatures, lead to an ice-off surface. The lake`s snow/ice cover during winter can act as a barrier, reducing the penetration of solar radiation through the water column and constraining the supply of nutrients into the lakes. During part of the springtime, the thinnest ice cover continues to act as a barrier to the supply of nutrients; however, it no longer acts as efficiently against solar radiation. Under these conditions, rapid and intense increases in UVB radiation can penetrate the ice layer (UVB radiation <310 nm) and reach the water column, as observed in a similar condition at Hoare Lake in McMurdo Dry Valleys, Antarctica [47]. In summer, the thin or absent ice cover exposes phytoplankton to a huge amount of solar radiation, including UVB, as well as periods of mixing and turbulence caused by wind and thermal stratification induced by temperature differences. During fall, phytoplankton experience periods of high mixing and turbulence, as well as thermal stratification, but UVB exposure is considerably lower. In addition to the ice/snow cover, UVR influx into the lakes is influenced by three other important parameters: i. total concentration of ozone column; ii. atmospheric optical depth, determined by aerosols, water vapor and cloud cover, surface albedo and solar zenith angle [48]; and iii. lake water turbidity.
For cloud cover in the Antarctic continent, a highly variable pattern has been reported. For example, on the Central Antarctic Plateau, it is nearly constant throughout the year, around 50–60% as inferred at the South Pole station [49], while in Maritime Antarctica it typically ranges between 80−90%. KGI is located in Maritime Antarctica, near the Drake Passage, a region characterized by the migration of two cyclonic systems per week, on average. The weather condition on the island is marked by a high cloud cover, which acts directly to reduce UV at ground level. In situ measurements at KGI from 2004 to 2011 have shown UV cloud modification factors (i.e., UV attenuation by clouds) between 30% and 70%, which are significantly higher than those found in mid-latitude sites, ranging from 15% to 32% [50]. Earlier studies conducted in Antarctica (at McMurdo and South Pole Stations) [48] have shown that UV attenuation by clouds is highly dependent on surface albedo and cloud optical depth, being more significant at low surface albedo due to multiple scattering within the cloud. Despite this, measured UVB radiation and ozone at KGI during October 1999 showed a correlation of ρ = −0.54 [51]. More recently, in situ data obtained in Punta Arenas, Chile/Southern Patagonia, also influenced by synoptic systems acting in the Drake Passage, provided a correlation between UVB/UVA (%) and effective Ozone (DU) of ρ = −0,92 [52]. These results demonstrate that regional cloudiness does not substantially impact the UVR levels which reach the surface.
In addition to the atmospheric filter, on ice-free aquatic surfaces, dissolved organic carbon (DOC), chromophoric dissolved organic matter (cDOM) [53], mineral suspended particles [8], and surface reflection derived from terrestrial allochthonous sources, may also act as attenuators of incoming solar UVR in the lacustrine environment [54]. According to Depauw et al. [55], when light enters the water column, it becomes partially polarized and is absorbed more strongly than in air in a wavelength-dependent manner. The blue light component, for example, is not strongly absorbed by water and can penetrate much deeper depending on the concentration of cDOM and suspended particles. In a study conducted in East Antarctic lakes [56], the effect of cDOM was found to be relatively low, as dense vegetation is absent in the catchments. According to the authors, proglacial lakes with transparent waterallow a substantial percentage of incident UVR (35–45%) to penetrate through lake ice, and in the absence of snow/ice cover, at least 10% of incident UVR can penetrate up to 3.5–6 m depth in ice-covered lakes, affecting living organisms. As satellite images and a local visit to Profound lake show, sediment input is very low, even during the seasonal peak of snowmelt water flowing into the lake. Therefore, the Fildes peninsula proglacial environment provides suitable conditions for the full development of diatoms.
Sediment core sampling
In Profound Lake, the sediment core was retrieved from the center of the lake (14 m depth) using a gravity core in March 2013. A 0.5 m sediment core was recovered in a plastic tube, sealed and kept refrigerated during transport to Brazil. In laboratory conditions, slices at 1 cm resolution were obtained using a disc saw at Federal Fluminense University (UFF)/Brazil. The subsamples were identified and sealed in plastic bags and stored in a freezer at −20°C until analysis.
In Geographensee lake, the sediment cores were retrieved from the deepest part of the lake, using a Russian corer in November 2019. A total of five sediment cores, each with 1 m in length and 5 cm in diameter, were sampled with a 10–20 cm overlap. The overlapping segments of sediment cores were compared visually by defining distinct marker layers and other prominent features of lithology. After the analyses using photogrammetry (Photoscan) and magnetic susceptibility measurements with the Geotek MSCL-XYZ system, five selected cores provided a continuous sedimentary sequence with a length of 4.24 m long, except for a few centimeters of the upper heavily watered layer lost during the sampling.
Diatoms analysis
A small amount (1g) of each subsample was prepared for diatom observation and identification using light microscopy (LM). For sample preparation, an adaptation of the method described in Crosta and Koç [57] was applied, in which the sediment was cleaned by adding 37% H_2_O_2_ and heating to 60°C for about 2 hours. The reaction was completed by the addition of 30% HCl, followed by rinsing with distilled water. Following digestion and centrifugation (three cycles of 2 minutes at 1200 rpm), the cleaned material was diluted with distilled water to avoid excessive concentrations of diatom valves during analysis. Slides containing the cleaned diatom material were prepared using Acrilex brand Colorless Stained Glass 500® as a mounting medium, employed as an alternative resin [58], and stored as permanent slides. Two slides were prepared for each subsample to perform qualitative analysis using a Nikon Eclipse TS100 microscope at 1,000x magnification. All diatoms were counted in random fields until at least 500 individual valves were recorded. Specimens representing more than half of a valve were counted as one.
Scanning electron microscopy (SEM) was used for the identification of selected diatom species. For this technique, aliquots of the oxidized suspensions were dried onto cover slips and mounted on aluminum stubs, which were carbon-coated using a Denton vacuum carbon evaporator equipped with a Desk carbon accessory. The samples were examined using a JEOL JSM-6510LV SEM operated at 1 kV situated in LABMEV (Chemistry Institute- UERJ). Morphological terminology followed Round, Crawford and Mann [59], Crawford and Likhoshway [60], Houk [61], Houk and Klee [62], and Zidarova, Kopalová and Van de Vijver [34].
The Profound lake sediment core was subsampled at 1 cm intervals, and only the upper 23 cm were analyzed, as below this depth, despite identical sample processing, the slides were too turbid to allow a correct taxonomic identification. In contrast, for Geographensee Lake, the core was subsampled at 3 cm intervals due to the limited amount of sediment available in each layer, and the entire core was analyzed.
After diatom counting and identification, the relative abundance of each taxon was calculated using the Mueller-Dombois and Ellenberg [63] method. The taxa were classified according to their frequency of occurrence in the samples as most frequent (present in > 50% of the samples), frequent (present in 10%–50% of the samples), and rare (present in < 10% of the samples and/or with a relative abundance less than 2% in the subsamples).
Sediment core dating and chronological model
A high-resolution gamma spectrometry system was employed using an extended energy range coaxial hyperpure germanium (HPGe) detector (model GX5021 – Canberra) with a relative efficiency of 50% and a energy resolution of 2.1 keV (FWHM) at the ^60^Co peak (1.33 MeV). This system was installed inside a very low-background lead (Pb) shielding with an average thickness of 15 cm, internally lined with pure copper (Cu). The detector efficiency curve for sediment samples was performed using a NIST-traceable liquid standard solution (serial number HV951) containing a cocktail of radionuclides. The radionuclides included in the standard were ^133^Ba, ^57^Co, ^139^Ce, ^85^Sr, ^137^Cs, ^54^Mn, ^88^Y, and ^65^Zn. The total counting time for each sample was 24 h. The main radionuclides measured were the ^210^Pb, ^226^Ra, and ^137^Cs.
The geochronological model used in this work was the Constant Rate of Supply (CRS). Chronology is based on the unsupported fraction of total ^210^Pb in each sediment layer, calculated as ^210^Pb_unsuported_ = ^210^Pb_total_–^210^Pb_supported_, where ^210^Pb_supported_ = ^226^Ra. In this model, mass sediment accumulation rates provide a measure of sedimentation that varies with depth. Sedimentation rates are determined by plotting the logarithm of excess ^210^Pb activities against cumulative dry sediment mass and calculating the sedimentation rates over intervals with a constant slope.
According to Appleby and Oldfield [64], the age of sediment layers can be calculated using , where A0 is the sum of total unsupported ^210^Pb activity measured in the sediment column, and Ax is the total unsupported ^210^Pb activity bellow depth x, while λ = ^210^Pb radioactive decay constant of (0.03114 yr^-1^).
For radiocarbon (¹⁴C) measurements, five bulk sediment samples and one sample of organic material (mosses) from Geographensee Lake core were submitted to Accelerator Mass Spectrometry (AMS) at DirectAMS Radiocarbon Lab (Bothell, Washington, USA) and Laboratorium Datowań (Kraków, Poland). Calibration and modeling of the obtained dates was performed in OxCal v. 4.2.4 [65] using the IntCal20 calibration curve [66], and the results are summarized in Verkulich et al. [67].
Ozone data and the accompanying datasets used for model calibration
To determine whether environmental variables and processes influence the variation in the relative abundance of diatoms, we used the following datasets: ozone, air temperature, cyclone energy, Antarctic Oscillation Index (AAO), cloud cover above KGI, winter snowfall, and wind speed, as summarized in S2 Table.
Statistical methods
Correlations between the relative abundance of diatoms from Profound Lake and environmental/climatic variables (see S2 Table) were calculated using Spearman’s rank-order correlation in STATISTICA 13.2 software. This non-parametric test evaluates the statistical dependence between two variables, assessing how well their relationship can be described by a monotonic function, with statistical significance set at a 95% confidence level (P < 0.05).
Following this initial correlation analysis, the diatom taxa that showed significant correlations with the measured ozone data (October mean total column ozone from the Faraday/Vernadsky station; 65°15′S, 64°16′W; 16 m a.s.l.) were selected to perform a multiple linear regression (MLR) analysis aimed at reconstructing ozone variability. The model is described by Equation 1, where coefficients X₁, X₂, X₃, and X_n_ represent the relative abundances of each diatom taxon:
To calibrate the MLR model, we used the relative abundances of the selected diatom assemblage from the 13 upper sediment layers, which cover the same period as the in-situ ozone measurements (1957–2013). The model’s performance was then evaluated by calculating a new Spearman’s correlation between the measured ozone values (Faraday/Vernadsky station) and those reconstructed from the relative abundance of the selected diatom assemblage. Finally, the calibrated model was applied to diatom assemblages from deeper sediment layers of the Profound Lake core to infer past ozone variability over time.
Results
Profound Lake dating using ^210^Pb for the 23 cm core top provided an age range spanning approximately 1911–2011 CE. This timescale, combined with the resolution of the core subsamples, allowed us to compare the ozone depletion period from 1980 to the present with the diatom species found in the core stratigraphy at decadal to semi-decadal resolution. This approach enables us to investigate the relationship between ozone depletion and diatom assemblages. In the uppermost section of the core, we identified thirty-two diatom taxa at the species or subspecies level. Half of these taxa are described as endemic to Antarctica and the Antarctica islands, whereas the remaining are cosmopolitan species occurring in Antarctica as well as other regions worldwide (i.e., Diatomella balfouriana Greville 1855 and Melosira varians C.Agardh 1827,which are also reported from Asia, Africa, North America, South America, and Europe) [34]. Among all identified species, only Aulacoseira glubokoyensis Oaquim, Moser, Evangelista & Van de Vijver 2017, the most abundant and dominant species in the investigated section of the lake, has so far been recorded exclusively in Profound Lake, KGI, and was first described by the authors in 2017 [68].
Over all the identified diatoms, thirteen taxa were classified as rare, as their frequency of occurrence was lower than 10% of the total samples and/or their relative abundance was below 2% in each subsample. The remaining nineteen taxa were classified as abundant and grouped into six categories (A–F) using Ward’s hierarchical clustering method (Fig 2), which reflects shared ecological preferences and similar responses to environmental drivers, such as wind stress, reductions in ice cover, nutrient inputs, and changes in light availability and UV exposure related to ozone variability.
Most abundant diatom taxa identified from the Profound Lake core.Bar chart indicates the relative abundance of the most frequent and frequent freshwater diatoms (relative abundance above 2%) grouped into six ecological clusters (A–F) using Ward’s hierarchical clustering. In the graph on the right, the number of taxa counted per sediment layer is shown in red, and the number of valves counted is shown in gray showing the variation in abundance and diversity of taxa along the corer.
For instance, 98% of the taxa grouped into the categories are benthic freshwater species typical of stable lake environments. The main exceptions are Melosira cf. varians and Tryblionella debilis Arnott ex O’Meara 1873 (not shown in Fig 2, as they were categorized as rare taxa with < 2% abundance), both of which are marine species. M. varians exhibited higher abundance peaks in the upper core layers, particularly after 2008 and around 1960. This pattern is consistent with an intensification of westerly wind stress inferred for the Southern Ocean during the December–February period [69] and with a decline in Bellingshausen Sea ice extent [70], suggesting a potential influence of sea spray on the lake environment, indicating that the occurrence of those taxa may reflect increased marine aerosol input associated with intensified westerly winds.
Among the freshwater taxa, several species such as Nitzschia cf. kleinteichiana Hamsher, Kopalová, Kociolek, Zidarova& Van de Vijver 2016; Psammothidium abundans (Manguin) Bukhtiyarova & Round 1996; Psammothidium confusoneglectum Kopalová, Zidarova & Van de Vijver 2016; Fragilaria cf. parva (Grunow) A. Tuji& D. M. Williams 2008; Humidophila tabellariaeformis (Krasske) R. L. Lowe et al. 2014; Pinnularia borealis Ehrenberg 1843 (34); and Gomphonema sp.—a cosmopolitan freshwater genus, commonly occurs in mats associated with submerged mosses and in surface lake sediments. These shaded microhabitats provide protection from excessive light intensity and UV radiation.
Diatom assemblage and the ozone reconstruction
After establishing a chronology of the diatoms relative abundance within the stratified layers of Profound Lake, we compared each taxon and assemblage group with ozone column concentrations for the corresponding years. Ozone data, expressed in Dobson Units (DU), were obtained from measurements at Faraday/Vernadsky, the closest ozone monitoring station to King George Island (located approximately 429 km away).
Analyzing both the individual species–ozone correlations (Fig 3) (numerical correlation values are provided S1 Table) under the light of Ward’s hierarchical clustering method, we identified four taxat hat exhibited statistically significant correlations (p < 0.05) with ozone concentration, despite belonging to distinct Ward’s categories: Gomphonema sp. (ρ = 0.73; p = 0.000; n = 13), Nitzschia cf. kleinteichiana (ρ = 0.64; p = 0.020; n = 13),Humidophila tabellariaeformis (ρ = −0.75; p = 0.000; n = 13), and Pinnularia borealis (ρ = 0.80; p = 0.000; n = 13).
Diatom microphotographs and Spearman correlations.Inventory of the abundant diatoms detected in the sediment core and their corresponding microphotographs. Bar color represents Spearman correlations (ρ) calculated between the relative abundance of diatom taxa (derived from dated sediment layers) and the corresponding ozone concentrations at the Faraday/Vernadsky Station, Antarctica. Ozone data refer to the October ozone column concentration. Yellow boxes highlight correlations that are statistically significant within a 95% confidence interval. White bars in the microphotographs indicate scale length: 10 µm for valve views in panels (a), (d), (e), (k), (n), (o), and (p); 2 µm for lateral views in panels (b) and (f); 1 µm for valve view in panel (c); 2 µm for valve views in panels (h), (i), (j), (q), and (r); and 5 µm for valve views in panels (g), (l), (m), and (s).
Even though these taxa (Gomphonema sp., Nitzschia cf*. kleinteichiana*, and Pinnularia borealis) fall in different categories, they reach very low or even zero relative abundance after 1981 up to the top of the core, following the same behavior of ozone variability. In contrast, Humidophila tabellariaeformis displays the opposite pattern, with increasing abundance during the period of ozone loss, which is characterized by higher UVB levels, suggesting a greater tolerance to elevated radiation.
However, changes in meteorological parameters could also, in principle, influence the distribution of these living diatoms in the lake, and their potential impacts should also be investigated. Accordingly, we tested the correlation between the diatom assemblages and concurrent meteorological/climatic parameters (including wind, cloudiness, cyclone energy, the AAO index, snowfall, and air temperature) (Table 1) during the same period of ozone depletion, to assess whether the grouped taxa responded exclusively to ozone variation or were also influenced by other environmental drivers.
Table 1: Diatoms set and the environmental correlations.
Among all tested parameters, only summer cyclone energy (50°–70°S) exhibited a statistically significant correlation (p < 0.05) with the diatom assemblage used for the ozone reconstruction. This finding reinforces the interpretation that the distribution of dominant benthic taxa in Profound Lake is closely linked to large-scale atmospheric processes.
Since the intensity and latitudinal position of summer cyclones are strongly modulated by the positive phase of the Southern Annular Mode (SAM) and by stratospheric ozone depletion [71], we interpret ozone variability as the primary driver influencing the diatom record, rather than the cyclone dynamics themselves. Based on this result, we applied a multiple regression model to the dataset to reconstruct ozone variability over time. The resulting multiple regression equation (Equation 2) is presented below:
where, O_r_: Reconstructed Ozone; Nk: relative abundance of Nitzschia cf. kleinteichiana; Pb: relative abundance of Pinnularia borealis; G: relative abundance of Gomphonema sp.; and Ht: relative abundance of Humidophila tabellariaeformis. The time interval covered by the reconstruction spans 1953–2013. In this case, the correlation between measured and modeled ozone was ρ = 0.95 (p = 0.00001) for 13 dated sediment strata (Fig 4).
Measured and modeled ozone concentration.Spearman correlation between modeled ozone concentrations based on the relative abundance of Gomphonema sp., Nitzschia cf. kleinteichiana, Humidophila tabellariaeformis, and Pinnularia borealis and measured October total ozone concentration at Faraday/Vernadsky Station.
The sediment core of Profound Lake allowed dating of the sediment strata back to approximately 1911, corresponding to a depth of 22 cm. In the reconstructed data, the onset of ozone depletion is observable from 1969 onwards. The minimum value occurs around the year 1992 (253 DU), whereas the in situ measurements indicate that the lowest ozone concentrations were recorded in 1991 (191 DU) and 2000 (193 DU). Our model also indicates a slight increase in ozone values, reaching a peak in 2006 (280 DU), which is close to the measured peak in 2002 (301 DU).
Considering that Profound lake and Geographensee lake are only 6.54 km apart, located within the same ice-free area, and therefore exposed to similar environmental, climatic, and geological conditions, we extended back the ozone reconstruction using the diatom database of Geographensee lake. In this case, we applied the multiple regression model developed for Profound lake to the same set of diatoms from the Geographensee lake core. The core spans the period from 613calyr B.P. to 7714 cal yr B.P (Fig 5). The long-term reconstructed ozone variability obtained using this above method shows that during the late-to-middle Holocene, reconstructed ozone concentrations were relatively stable, ranging from 305 to 384 DU, similar to ozone levels observed in the decades preceding the CFCs epoch. According to our reconstruction, the ozone depletion event in Antarctica since the 80´s decade is unprecedented over the last ~7.7 cal kyr B.P.
Ozone reconstruction model from the modern to the middleHolocene period based on diatom assemblages.Blue line is the total ozone (DU) measured ar the Faraday/Vernadsky Station, Antarctica. The black line is ozone reconstructed based on four diatom taxa (Gomphonema sp., Humidophila tabellariaeformis, Pinnularia borealis, and Nitzschia cf. kleinteichiana) from Profound lake (left side) and Geographensee lake (right side). Bar charts refer to the relative abundance of diatoms.
Discussion
Although no direct relationship between stratospheric ozone variations and lake diatoms has been established in the literature, the relationship between UVR and diatoms is widely reported and well established for some species, with effects varying at the intra- and interspecies levels (i.e., Rech et al. [72], Fouqueray et al. [73], Roy et al. [74]). In the Profound lake sediment core, most identified taxa are benthic diatoms, which commonly possess a raphe—a slit in the frustule that enables motility [27]. Their benthic lifestyle is supported by physiological adaptations to low light intensities near the substrate, while the raphe represents a morphological trait associated with this mode of life.
From the nineteen most abundant taxa, only four showed significant correlations with ozone variability: i. Gomphonema sp., a cosmopolitan freshwater genus that commonly occurs in mats associated with submerged mosses at the bottom of lake sediments. These shaded microhabitat provide protection from excessive light and UV radiation. Therefore, higher ozone concentrations — which correspond to reduced UV exposure — may favor the proliferation of Gomphonema sp., consistent with its adaptation to low-light environments; ii. Pinnularia borealis, a freshwater and terrestrial species commonly found in mosses, water effluents, and permanent aquatic or wet (or humid) habitats [34], is described as a live motile species that is rarely found in water bodies. As it was observed at high relative abundance in 36% of the subsamples, we believe that it had an autochthonous origin; iii. Nitzschia cf. Kleinteichiana, a freshwater species endemic to Maritime Antarctica with broad ecological amplitude, is often found forming large populations in lakes, ponds and melting water streams; iv. Humidophila tabellariaeformis is the only diatom whose relative abundance varied inversely with ozone levels, suggesting a possible adaptation to elevated UVB radiation. As a freshwater and cosmopolitan species [75], it inhabits cyanobacterial mats at the bottoms of ponds and within moss vegetation [34], which may provide cellular protection and facilitate acclimatization. Currently, no studies in the literature have evaluated the impact of UVR on any of these taxa, nor have the protective and repair mechanisms employed by these cells under varying radiation conditions been described, as has been done for other species such as Coscinodiscus granii [27] and Coscinodiscus wailesii [76].
Previous analysis of westerly winds around Antarctica supports the hypothesis that their intensification after the 80´s decade is partly related to the stratospheric ozone depletion, through a complex interaction involving stratospheric cooling and the poleward shift of the westerly winds, especially during summer [77], despite the influence of the positive phase of the SAM [78]. The southwards shift of the westerlies has significant implications for storm tracks and, consequently, for the exchange of air masses in the Maritime Antarctica region. Therefore, the period corresponding to ozone depletion in Antarctica is also characterized by important changes in regional synoptical conditions. Increases in westerly wind intensity, air temperature [79], and cyclonic energy [80] have also been reported for the same period. These changes in meteorological parameterscould, in principle, influence the distribution of living diatom in the lake, and their potential impact was also investigated (summarized in Table I).
These synoptic-scale atmospheric changes are particularly relevant for lacustrine ecosystems on King George Island, where the climate is strongly influenced by marine air masses and exhibits high seasonal variability. Frequent advection of moist and relatively warm air from the Southern Pacific and Southern Atlantic promotes a wide range of precipitation types and pronounced air temperature fluctuations throughout the year. According to the detailed air temperature compilation by Poelking et al. [81], mean temperatures reach the melting point during most months, except for June, July, and August. As a result, lakes commonly alternate between fully frozen, partially ice-covered, and ice-free conditions. This variability directly affects ice-cover duration, light penetration, water column stability, and the extent of benthic habitats. Consequently, interannual anomalies in atmospheric circulation and temperature associated with ozone depletion may indirectly modulate diatom community structure by altering habitat availability and exposure to UV radiation, particularly for benthic and motile taxa adapted to low-light environments.
The diatom-based ozone reconstruction from Profound Lake (Fig 6H) reveals a century-long pattern of variability since 1911, reflecting both natural and anthropogenic forcings acting over the Antarctic Peninsula. Volcanic eruptions, both local and global, (Fig 6E, F), likely contributed to episodic ozone depletion through the injection of halogenated species (hydrogen chloride - HCl, bromine monoxide- BrO and iodine monoxide- IO) into the stratosphere. These gases, together with sulfate aerosols produced during eruptions, enhance the heterogeneous reactions responsible for ozone loss by increasing the reactive surface area available in the stratospheric aerosol layer [5,82].
Ozone forcings and reconstruction models.(a) Timeline of equivalent effective chlorine (EESC) (adapted from WMO, 2018); (b) Total Solar Irradiance (TSI) reconstruction by Krivova, Vieira and Solanki (2010) [85] (c) UV-B, 280 nm from NRLSSI2; (d) Perturbations in the mean annual optical depth (visual, λ = 0.55μm) for both hemispheres resulting from volcanic eruptions in the Southern Hemisphere adapted from Zielinski [84]; (e) confirmed volcanic eruptions data in Antarctica, obtained from the Global Volcanism Program; (f) Confirmed Global Volcanic Eruptions data, with VEI > 4 (source: Global Volcanism Program, https://volcano.si.edu/); (g) number of diatom taxa found in Profound Lake; (h) Blue line shows total ozone (DU) from Faraday/Vernadsky station, Antarctica; black line shows reconstructed ozone (this issue); (i) Ozone layer depth (DU) reconstructions, adapted from Chen et al. [86], black line shows TSI and red line the Solar Spectra Irradiance (SSI) in the 280-450nm; (j) Antarctica UVR reconstruction based on mycosporine-like amino acids as proxy [87]. Gray box highlights the period when ozone concentrations start to decrease for previously unseen values.
Solar variability (Fig 6B–C) also influences stratospheric ozone through changes in ultraviolet irradiance, especially in the UVC band (<242 nm) responsible for the photolysis of molecular oxygen. However, our record does not show a consistent link between ozone concentrations and solar cycles at decadal to centennial scales. Although the 100-year Gleissberg cycle shows a decline after the 1970s, no corresponding oscillations were observed in the reconstructed ozone trend [83].
Conversely, volcanic activity after the 1960s coincides with pronounced anomalies in the ozone reconstruction. Significant eruptions—such as Agung (1963), Fernandina (1968), Fuego (1974), El Chichón (1982), and Pinatubo (1991)—produced notable increases in stratospheric optical depth (Fig 6D), which is consistent with periods of enhanced ozone depletion in the Southern Hemisphere [84]. From the 1980s onward, both measured and modeled ozone data reveal an abrupt decline, primarily attributed to the increasing emission of chlorofluorocarbons (CFCs) (Fig 6A). These compounds release reactive chlorine radicals (Cl and ClO) under intense UV radiation, further accelerating ozone loss processes in the Antarctic stratosphere [83].
Overall, the agreement between the diatom-inferred ozone variations and known atmospheric forcings supports the sensitivity of the Profound Lake record to large-scale stratospheric processes. The strong association with summer cyclone energy and volcanic perturbations underscores the potential of these lacustrine systems as biological archives for tracking the legacy of ozone depletion over the Antarctic Peninsula.
Our diatom-based ozone reconstruction captures historical trends with higher temporal resolution than previous models. Compared with the TSI/SSI-based reconstruction by Chen et al. [86] (Fig 6I), it shows a closer fit to measured ozone, likely because it focuses on the Maritime Antarctic rather than a broad latitudinal band. Similarly, compared with UV-B reconstructions from the Ross Sea using photoprotective pigments [87], the diatom-based reconstruction resolves finer-scale changes. Minor temporal shifts can be attributed to dating uncertainties or delayed diatom responses in high-latitude sediments. Overall, these results underscore the robustness and higher resolution of diatom-based reconstructions for understanding past ozone variability.
Also, the reconstruction exhibits patterns similar to those derived from major photochemical drivers such as SSI, as suggested by Chen et al. [86], with both depicting the dramatic ozone loss after the 1980s. In comparison with measured ozone, diatoms present a better fit, whereas the SSI reconstruction shows residuals of ~30 DU, likely reflecting the broader latitudinal scale of the SSI model versus the localized focus of our study. Minor temporal shifts in the diatom record may result from: i) dating errors of a few years, which could artificially enhance correlation with ozone; ii) temporal smoothing due to the sediment core resolution, which preserves an average of 5 years of deposition per 1 cm interval, reaching up to 9 years in some layers; and iii) delayed diatom responses in recent sediments, a pattern previously observed in deep, high-latitude lakes with extensive ice cover [88].
A second related reconstruction in Antarctica, in the Ross Sea area (between Victoria Land and Marie Byrd Land) (Fig 6J), employed algae by-products (MAAs, scytonemin, and total cyanobacterial carotenoids, including canthaxanthin, echinenone, and zeaxanthin) as photo-protective proxies for UV-B radiation [87]. While these pigment-based reconstructions show good agreement with higher UV-B levels in recent times, our diatom-based reconstruction provides higher-resolution insights into ozone variability, further highlighting the advantage of using diatoms to reconstruct past stratospheric ozone dynamics.
Conclusion
Stratospheric ozone depletion over Antarctica during the modern era and its effects over the biota, biogeochemical processes and the regional climate has been well documented. Since 1956, Antarctic ozone measurements using a Dobson Spectrophotometer have varied from 320 DU to nearly 100 DU, representing a unique scenario unprecedented by man. This makes the Antarctic biosphere and lower atmosphere a natural laboratory for studying interactions between living organisms and biogeochemical cycles under enhanced UV radiation, alongside multiple environmental drivers. Lake diatoms, due to their high sensitivity to environmental changes, emerge as a powerful proxy for this unique phenomenon since the onset of the industrial era. Analysis of well-preserved and precisely dated sediment profiles enabled the reconstruction of ozone depletion history from mid-to-late Holocene till the present time. Extending this approach to the past few thousand years we estimated a relatively uniform behavior for the ozone concentrations never exceeding the interval 290–320 DU, despite the fact of an existing volcanic active period in Late Holocene for Antarctica and the finding that no compelling evidence for long-term variations in total solar activity is recorded on millennial timescales. For the modern epoch our results corroborate other reconstruction models derived from UV-induced photoprotective molecules preserved in sediment layers and others based solely on solar irradiance variability. Considering the whole ozone reconstruction history covering the last 5,000 years, we concluded that anthropogenically induced ozone depletion is likely unprecedented over this timescale, representing one of the most severe human-induced threats to natural polar ecosystems during mid-to-late Holocene.
Supporting information
S1 TableSpearman correlation between the relative abundance of the most abundant diatoms set and the environmental data.Blue and gray boxes corresponds to ρ and p values, respectively. Values in bold are those that present significant statistical correlations.(DOCX)
S2 TableOzone data and other accompanying databases used for model calibration.(DOCX)
S1 FigChronological age-depth model and sedimentation rate from Profound Lake.(a) 210 Pb excess observed along the Profound lake core; (b) Profound lake layers age based on the model CSR showing the minimum and maximum age of each sediment layers; (c) Sedimentation rate from profound Lake/ KGI; (d) Sedimentation rate from Lake long/KGI.(DOCX)
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Chipperfield MP, Dhomse SS, Feng W, Mc Kenzie RL, Velders GJM, Pyle JA. Quantifying the ozone and ultraviolet benefits already achieved by the Montreal Protocol. Nat Commun. 2015;6:7233. doi: 10.1038/ncomms 8233 26011106 PMC 4455099 · doi ↗ · pubmed ↗
- 2Farman JC, Gardiner BG, Shanklin JD. Large losses of total ozone in Antarctica reveal seasonal Cl Ox/N Ox interaction. Nature. 1985;315(6016):207–10. doi: 10.1038/315207 a 0 · doi ↗
- 3Molina MJ, Rowland FS. Stratospheric sink for chlorofluoromethanes: chlorine atom-catalysed destruction of ozone. Nature. 1974;249(5460):810–2. doi: 10.1038/249810 a 0 · doi ↗
- 4Stolarski RS, Cicerone RJ. Stratospheric chlorine: a possible sink for ozone. Can J Chem. 1974;52(8):1610–5.
- 5WMO (World Meteorological Organization). Scientific Assessment of Ozone Depletion: 2018, Global Ozone Research and Monitoring Project–Report No. 58. Geneva, Switzerland; 2018. pp. 588.
- 6Zhu Y, Toon OB, Kinnison D, Harvey VL, Mills MJ, Bardeen CG. Stratospheric aerosols, polar stratospheric clouds, and polar ozone depletion after the Mount Calbuco eruption in 2015. J Geophys Res Atmos. 2018;123(21):12308–31.
- 7Zuev VV, Zueva NE, Savelieva ES, Gerasimov VV. The Antarctic ozone depletion caused by Erebus volcano gas emissions. Atmospheric Environment. 2015;122:393–9.
- 8Bancroft BA, Baker NJ, Blaustein AR. Effects of UVB radiation on marine and freshwater organisms: a synthesis through meta-analysis. Ecol Lett. 2007;10(4):332–45. doi: 10.1111/j.1461-0248.2007.01022.x 17355571 · doi ↗ · pubmed ↗