Antarctic ice-free areas contain lakes and ponds that have interesting limnological features and are of wide global significance as early warning indicators of climatic and environmental change. However, most limnological and paleolimnological studies in continental Antarctica are limited to certain regions. There are several ice-free areas in Victoria Land that have not yet been studied well. There is therefore a need to extend limnological studies in space and time to understand how different geological and climatic features affect the composition and biological activity of freshwater communities. With the aim of contributing to a better limnological characterization of Victoria Land, this paper reports data on sedimentary pigments (used to identify the main algal taxa) obtained through a methodology that is more sensitive and selective than that of previous studies. Analyses were extended to 48 water bodies in ice-free areas with differing lithology, latitude, and altitude, and with different morphometry and physical, chemical, and biological characteristics in order to identify environmental factors affecting the distribution and composition of freshwater autotrophic communities. A wider knowledge of lakes in a limnologically important region of Antarctica was obtained. Cyanophyta was found to be the most important algal group, followed by Chlorophyta and Bacillariophyta, whereas latitude and altitude are the main factors affecting pigment distribution.
Introduction
Antarctic ice-free areas contain several lakes and ponds that have interesting limnological features and are of wide global significance, especially as refugia for unique species and as early warning indicators of climatic and environmental changes (Vincent and Laybourn-Parry, 2008). However, most limnological and paleolimnological studies in continental Antarctica are limited to certain regions such as the Larsemann Hills (e.g., Hodgson et al., 2001, 2006; Squier et al., 2004, 2005; Verleyen et al., 2004) and Vestfold Hills (Fulford-Smith and Sikes, 1996; Laybourn-Parry et al., 2002; Henshaw and Laybourn-Parry, 2002). Limnological research in ice-free areas of Victoria Land (from 79°S in Darwin Glacier to 71°S at Cape Adare) began about 50 years ago, focusing especially on the McMurdo Dry Valleys (76.50°–78.00°S and 160.00°–165.00°E) (e.g., Priscu, 1998). The latter is the largest ice-free area in Antarctica (4800 km2), with characteristics typical of cold deserts (precipitation <100 mm water equivalent/yr, mean annual temperatures of −16 to −21 °C and humidity <50%). As shown by the results of short-term (e.g., Green et al., 1989; Matsumoto, 1993; Webster et al., 1994; Spiegel and Priscu, 1998) and long-term investigations (e.g., Green and Friedmann, 1993; Priscu, 1998), streams and lakes in the McMurdo Dry Valleys are the sites of high biological activity during summer. However, there is a need to extend limnological studies in space and time in order to understand how different geological and climatic features throughout Victoria Land affect the composition and biological activity of freshwater communities. Although most lakes in the region have a perennial ice cover (3–5 m thick) with reduced radiant energy, gas exchanges, and productivity, several water bodies in the coastal ice-free areas have no ice cover during summer and contain simple truncated food webs with abundant bacteria, algae, ciliated and flagellated protozoans, and rotifers (Laybourn-Parry, 2009). The benthos is dominated by extensive microbial mats (mainly consisting of filamentous cyanobacteria, pennate diatoms, and green algae) of different morphology probably determined by local environmental conditions (Wharton et al., 1983).
Limnological surveys in northern Victoria Land were first completed in the 1990s, after the establishment of the Italian “Mario Zucchelli” Antarctic Station (MZS) at Terra Nova Bay (Ross Sea). The main morphometric, hydrochemical, and edaphic features of several lakes and ponds located near MZS were reported by Fanzutti et al. (1989), Guilizzoni et al. (1992), Libera (1993), and Baudo et al. (2000). More recently, the physicochemical characteristics of lacustrine waters and sediments were reported by Borghini and Bargagli (2004, 2005), Borghini et al. (2007, 2008), and Malandrino et al. (2009). In general, these studies show that the water chemistry largely reflects that of seawater; additional and alternative sources of solutes were only found for NO3−, SO42−, K+, and Ca2+. Ion concentrations are highly variable in space and time, and maximum values are usually measured in water samples collected at the end of summer from lakes with no ice cover (i.e. those with enhanced drainage and water evaporation; Borghini et al., 2008). Sedimentary concentrations of total carbon (TC), total nitrogen (TN), and S are very low, except in water bodies affected by the presence of seabirds. Biological investigations have been performed in lakes at Edmonson Point, Gondwana, Inexpressible Is., Tarn Flat, and the Northern Foothills (e.g., Broady, 1987; Guilizzoni et al., 1989; Andreoli et al., 1992, 1996; Fumanti et al., 1992, 1995, 1997; La Rocca et al., 1996; Cavacini, 1999; Cavacini and Fumanti, 2005), and microscopy-based studies have reported more than 100 taxa (not all at the species level). Phytoplankton communities are mainly characterized by Cyanophyta, Cryptophyta, Crysophyta, Chlorophyta and Bacillariophyta. The most abundant genera in microbial mats are Oscillatoria, Nostoc, Leptolyngbya, Chroococcus, Phormidium, Calothrix, Schizothrix (Cyanophyta), Navicula, Luticola, Hantzschia, Diadesmis, Pinnularia (Bacillariophyta), and Kentrosphaera, Pleurococcus, Binuclearia (Chlorophyta). Although microscopy studies are very informative, they require time-consuming sample preparation and counting, and the identification of small cells is often problematic. Recent studies on pigments from lacustrine sediments and microbial mats collected in East Antarctica (Hodgson et al., 2001, 2004a, 2005; Squier et al., 2002, 2005; Verleyen et al., 2005) show that this approach allows the identification of temporal changes in the photoautotrophic community composition and paleoenvironmental conditions. We therefore began a general survey on the water geochemistry of and photosynthetic pigments in lakes and ponds of Victoria Land with the aim of creating a database of present-day physical, chemical, and biological characteristics for future assessment of possible variations due to local climatic and/or environmental changes. Preliminary results (Borghini et al., 2007) showed that despite the low taxonomic resolution, high performance liquid chromatography (HPLC) analysis of sedimentary pigments is a valuable tool for identifying the dominant taxa in Antarctic freshwater ecosystems. With the aim of contributing to a better characterization of sedimentary pigments in Victoria Land, this paper reports data obtained using an HPLC methodology that is more sensitive and selective than that used in previous studies. Analyses were completed on 48 water bodies with different morphometry and physical, chemical, and biological characteristics and located in ice-free areas of differing lithology, latitude, and altitude in order to identify environmental factors affecting the distribution and composition of freshwater autotrophic communities.
Materials and Methods
Field surveys and sampling were performed during four Italian Antarctic Expeditions (austral summer 2001/2002–2005/2006) in lakes and ponds located in 19 ice-free areas between Kar Plateau (76.91°S, 162.54°E) and Helm Point (72.13°S, 170.15°E; Fig. 1). All lakes were endorheic (with no surface outflow), wet-based (not frozen to their bed in winter), and proglacial (i.e. they formed as a result of glacial retreat after the Last Glacial Maximum, when Victoria Land coasts were covered by a marine-based ice sheet). Their main characteristics are summarized in Table 1. The names of most lakes are not official and, whenever possible, we adopted the nomenclature previously used by other researchers (e.g., Guilizzoni et al., 1989). The general features of each ice-free area and its lakes were recorded during sampling; a scale was used to express ice cover conditions (no ice, partial cover, perennial cover) and the number of seabirds nesting nearby (no birds, few, several, many).
TABLE 1
Location of sampled lakes, their catchment lithology, distance from the sea (m), altitude a.s.l. (m), estimated lake surface (m2), presence of birds and ice-cover (at the time of sampling).
Superficial sediments (0–3 cm, at a fixed water depth of 20 cm) were collected from different points in the littoral zone. Samples were brought back to MZS station and stored at −20 °C for transport to Italy.
About 2 g of fresh sediments were extracted at low temperature using 5 mL acetone (90%, chromatography grade), a vortex, and a sonicator. They were then stored for 1 hour at −20 °C. These operations were repeated three times, and the extracts were centrifuged and filtered at 0.45 µm. Ammonium acetate 0.1 M (10% of the sample) was added to the sample just before analysis. Throughout the extraction procedure, samples were protected from light and high temperatures. The extracts were analyzed by Liquid Chromatography–Mass Spectrometry with an Atmospheric Pressure Chemical Ionization source coupled with a Photodiode Array Detector (APCI LC-MS/PDA). Reversed phase columns (two Spherisorb ODS2 Hypersil, 150 × 4.6 mm ID, 5 µm particle size equipped with ODS2 pre-column) were used along with a solvent system (Table 2) and slightly modified gradients (10 mM ammonium acetate, 50 µL of sample) according to the method described in Pinckney et al. (1996). LC-MS analysis was performed using a Thermo system comprising a Finnigan surveyor autosampler, a MS pump, and a Finnigan LTQ. APCI LC-MS analysis was performed in positive ion mode, and MS instrument settings were as follows: capillary temperature of 250 °C, APCI vaporizer temperature of 350 °C, discharge current of 5.5 µA, discharge voltage of 4 kV, sheath gas flow rate of 40 arbitrary units (a.u.), auxiliary gas flow rate of 14 a.u., and sweep gas flow rate of 0 a.u.
TABLE 2
High Performance Liquid Chromatography (HPLC) gradient method utilized for photosynthetic pigment analysis.
Pigments and more weakly acidic MAAs (palythenic acid, palythene, palythine) were detected at their maximum wavelength and characteristic m/z. Peaks were identified by their absorption and characteristic MH+ and MS2 spectra, and their elution order through comparison with literature data (e.g., Rozema et al., 2002; Whitehead and Hedges, 2002; Yuan et al., 2009). Their concentrations were determined by comparing HPLC peak areas with those of standard solutions prepared using commercially available purified pigments from the International Agency for 14C Determination, VKI in Hoersholm, Denmark (pheophytin a, pheophorbide a, chlorophyllide a, chl c2, chl c3, divinyl chlorophyll a, lutein, peridinin, fucoxanthin, violaxanthin, canthaxanthin, zeaxanthin, alloxanthin, neoxanthin, 19′-butanohyloxyfucoxanthin, echinenone, prasinoxanthin, antheraxanthin, diadinoxanthin, 19′-hexanoyloxyfucoxanthin, lycopene, myxoxanthophyll), and from Sigma Aldrich (chl a and b, ß,å- and ß,ß-car). Concentrations of unavailable standard molecules were estimated from the peak area, assuming a specific extinction coefficient for each compound, as reported in the literature (Hurley and Watras, 1991; Villanueva et al., 1994; Jeffrey et al., 1997). When extinction coefficients were not reported or when no structurally or spectrally similar pigments existed, the coefficient of ß,ß-car was used (Jeffrey et al., 1997). Data quality was checked every six samples by analyzing blanks (acetone diluted with 10% acetic ammonium), which were always below the detection limit. Replicate and duplicate samples were analyzed, and the relative standard deviation (SD) was usually <20%. Results were expressed as µg g−1 TOC, because comparison between long-term monitoring data on lake plankton and the resulting varved fossil record indicates that this unit of measurement most accurately captures variations in algal abundance and community composition (Leavitt et al., 1997).
TOC in sediments was determined through a modified Walkley-Black titration method. TC and TN were analyzed with an elemental analyzer (2400 Series II, Perkin-Elmer), whereas P and S were determined by atomic emission spectrometry (Optima 5000 DV, Perkin-Elmer). Data quality was checked for each batch of samples by analyzing blanks (MQ waters were used for dilutions) and by simultaneous analysis of Standard Reference Materials with certified values. Replicate and duplicate samples were run daily, and the relative standard deviation between duplicates was always ≤1%.
STATISTICS
Spearman rank correlation was used to assess correlations between variables. Non-metric multidimensional scaling (n-MDS) was used to produce two-dimensional ordinations of the rank order of similarities (Anderson and Underwood, 1997) in order to compare the lakes on the basis of pigment distribution alone: a matrix of similarities between each pair of these samples was calculated using the Bray-Curtis similarity coefficient (Bray and Curtis, 1957). As a measure of the goodness of fit of a two-dimensional nMDS ordination space, we used the STRESS index (Legendre and Legendre, 1998). We adopted a scaling for which a STRESS < 0.2 indicates good performance in terms of the ability of the two-dimensional ordination to account for the relative position of samples in the multivariate space. Prior to multivariate analysis, all data were log-transformed to reduce or remove skewness. Redundancy Analysis (RDA; ter Braak, 1986; Legendre and Legendre, 1998) was used to investigate environmental factors affecting the distribution of pigments (Colacevich et al., 2009). Binary, categorical variables were introduced as dummy variates (0,1). Categorical variable with multiple, ordered levels (e.g., Ice Cover: 0, 1, and 2), were entered as dummy factors coded by a uniform sequence of integers. All analyses were performed using R version 2.8.1 (R Development Core Team, 2006: http://www.r-project.org). In particular, multivariate analyses were performed using the vegan package (Oksanen et al., 2006).
Results
S, TP, TOC, and TC concentrations in superficial sediments from Victoria Land lakes (Table 3) varied by one order of magnitude.
TABLE 3
Water content (W), TN, TC and TOC (in %), S, and TP (expressed in µg g−1 dry weight) in superficial lake sediments; N, number of lakes sampled at each site.
A total of 132 compounds (including pigments, UV compounds, and MAAs; Table 4) were detected in superficial sediments. Chl a, pheophytin a, and native and reduced scytonemins were the most widespread pigments. Other very common compounds were chl b, ß,ß-car, and some UV photoprotective compounds such as palythenic acid (e.g., Bandaranayake, 1998; Carreto et al., 2005 and references therein). Among algal marker pigments, the most frequent were echinenone, fucoxanthin, lutein, canthaxanthin, and myxoxanthophyll. The highest pigment concentrations were measured in sediments from lakes at Edmonson Point, Edmonson Point North, Gondwana, and Luther Peak (Table 5). Concentrations of chl a were significantly correlated (p < 0.01) with those of chl b (r = 0.77), lutein (r = 0.67), neoxanthin (r = 0.51), echinenone (r = 0.63), myxoxanthophyll (r = 0.53), zeaxanthin (r = 0.44), fucoxanthin (r = 0.55), and bacteriochlorophyll a (r = 0.40).
TABLE 4
Summary of pigment and MAA features determined by High Performance Liquid Chromatography–Photodiode Array Detector–Mass Spectrometry (HPLC-PDA-MS) in sediments.
TABLE 5
Average concentrations (in µg g TOC−1) of major pigments in Victoria Land lakes. The epimers (of chlorophylls, pyropheophytins, pheophytins, and carotenoids) were summed with the main isomers. The reduced scytonemin (Scyt red) includes the sum of reduced scytonemin and its derivatives. Σ-car, sum of all unidentified carotenoids; Σ-UV, sum of all unidentified UV photoprotective compounds, quantified using the Scytonemin extinction coefficient. Chlide a, chlorophyllide a; Phide a, Pheophorbide a; Fuco, fucoxanthin; Neo, neoxanthin; Viola, violaxanthin; Myxo, Myxoxanthophyll; Lut, lutein; Zea, zeaxanthin; Cantha, canthaxanthin; Bchl a, bacteriochlorophyll a; Chl b, chlorophyll b, Echi, echinenone; chl a, chlorophyll a; Phytin, pheophytin; ß,ß-car, ß,ß-carotene).
No significant (p < 0.01) correlation was found between water nutrients and pigments, whereas cyanobacterial markers (myxoxanthophyll, cantaxanthin, and echinenone) and bacteriochlorophyll a were correlated with sedimentary nutrients. All marker pigments were negatively related to latitude (chl a, chl b, lutein significantly) whereas only scytonemins were significantly and positively correlated to latitude.
Non-metric multidimensional scaling ordination (n-MDS) based on the presence-absence matrix of pigments and UV photoprotective compounds was used to identify the most similar lakes. Figure 2 shows that Lake 14, Lake 15, and Pond E at Edmonson Point, Lago Pantano at Edmonson Point North, Lagoon at Inexpressible Is., Pozza Eneide in the Northern Foothills, the two lakes at Gondwana and Luther Peak, and the lakes at Helm Point, Crater Cirque, and Depot Is. are very similar and form a cluster (in the circle). A posteriori data inspection revealed that they have the highest number of pigments, many of which are present in these lakes alone. In general, these lakes were characterized by the absence of any ice cover, and/or were located near the seashore and/or penguin rookeries and nesting seabirds. Another approach for interpreting the n-MDS map is to take two very distant objects and attempt to interpret the dimensions. For example, Lake G and Lake F on Inexpressible Is. and Lake 5 at Andersson Ridge differed most from the other lakes. They were characterized by a very low number of pigments.
RDA was used to study the distribution of pigments as a function of environmental characteristics (distance from the sea, latitude, altitude, ice cover, bird presence and catchment lithology) and the water chemistry (conductivity, Cl−, NO3−, SO42−, Na+, K+, Mg2+, Ca2+, total Nitrogen and Phosphorous, TN and TP, and Silicates) and sediments (TN, TP, TOC and S). Only the analysis conducted using the environmental variables as explanatory matrix was statistically significant. The permutation test performed on all eigenvalues indicated that the above explanatory variables accounted for a significant (999 permutations, p< 0.05) portion of pigment distribution variations. Overall, the first and second RDA axes accounted for 65% and 19% of the explained variance respectively (Fig. 3); in particular, altitude (along the first axis) and latitude (along the second axis) were the two most important variables accounting for the distribution of pigments.
Discussion
Data from two lakes (Lake 14 and Lake 15) at Edmonson Point revealed a sedimentation rate of 0.033 and 0.039 cm year−1, respectively (unpublished data), indicating that the each cm of sediment has accumulated over a period of 30 and 25 years, respectively. These sedimentation rates are intermediate between those reported in Progress Lake (0.14 cm year−1) and Lake Reid (0.23 cm year−1) in Larsemann Hills Region (Hodgson et al., 2001, 2005; Squier et al., 2005) and the sedimentation rate of Lake Hoare in the Dry Valleys (∼0.015 cm year−1; Doran et al., 1999). The latter is a perennially ice-covered lake, whereas the Larsemann Hills lakes have a longer ice-free period than our studied lakes due to the milder climate conditions of the region. Thus, even if uncertainty about sedimentation rates at the other sampling sites made estimation of annual pigment fluxes impossible, it is plausible to hypothesize that generally the superficial 3 cm samples have accumulated over a period of some tens of years.
S, TP, TOC, and TC concentrations in superficial sediments, as well as average TOC/TN and TOC/S ratios (2.7 and 8.8, respectively), were in the same range as those determined in the previous investigation of Victoria Land lakes (Borghini et al., 2007), whereas a much higher number of compounds were detected in superficial sediments.
Assuming chl a as an indicator of overall photoautotrophic biomass; chl b, neoxanthin, violaxanthin, and lutein as indicators of Chlorophyta; echinenone, myxoxanthophyll, zeaxanthin, and canthaxanthin as indicators of cyanobacteria; and fucoxanthin as an indicator of Bacillariophyta and Crysophyta, results indicate that Cyanophyta is the most abundant algal group in Victoria Land lakes. The presence of lutein and fucoxanthin in most sediments indicates that Chlorophyta and diatoms are also widespread, along with anaerobic bacteria (as indicated by bacteriochlorophylls and their degradation products). The correlations of chl a with chl b and lutein and echinenone and myxoxanthophyll indicate that primary production is due especially to Chlorophyta and Cyanobacteria.
Chlorophylls are susceptible to a variety of transformation reactions. Many attempts have been made to identify specific sources of particular chl a products. While studying chl a oxidation products in Kirisjes Pond (Larsemann Hills), Walker et al. (2002) found a relationship between the extent of chl oxidation and the residence time of the pigment in an oxygenated water column. Chlorine steryl esters (CSEs) are usually regarded as specific markers of grazing activity, and Squier et al. (2005) suggested that their occurrence in a sediment core from Progress Lake (East Antarctica) could be used to infer a marked change in community composition during the last interglacial period when the pressure of grazing had a significant impact on the phytoplankton community. No CSEs were found in the superficial sediments of the studied lakes, indicating that this area is characterized by truncated food webs with low grazing rates.
In general, pigment profiles and concentrations in the surface sediments of Victoria Land lakes were in the same range as those reported for several lakes in the Larsemann Hills (Squier et al., 2002; 2005), and the detected taxa corresponded to those reported in previous microscopy studies on cyanobacterial and algal mats from some of the sampled lakes (e.g., Broady, 1987; Fumanti et al., 1997; Cavacini and Fumanti, 2005).
Multivariate (n-MDS) analysis suggests that there are large-scale gradients in the diversity of pigments, and altitude and latitude were the two most important variables accounting for the distribution of pigments. RDA analysis partly confirms the results of Discriminant Function and Canonical Correspondence analyses completed in a previous study (Borghini et al., 2008): these analyses indicate that the distribution of pigments in Victoria Land lakes depends mainly on distance from the sea and the nutrient status. Trend of decrease in species diversity with latitude has been attributed to the harshness of the environment such as the period of ice cover and light intensity and to geographical isolation (Jones, 1996). The positive relationship between scytonemins and latitude seems to indicate higher UV levels at higher latitude. Recently, light availability was found to be a major controlling factor of lake productivity in nutrient-poor lake ecosystems (Karlsson et al., 2009). The distance from the sea could influence microbial mat composition and structure because the lakes located close to the sea had a shorter ice cover period than the plateau lakes, likely due to their altitude and salt content causing a delay in freezing and earlier melt out; moreover, they are generally nutrient enriched from the sea. The Victoria Land (from Cape Adare, 71°S, to the southern end of Ross Island, 86°S) contains the most extensive coastal gradient in Antarctica and includes a variety of habitats. Important environmental factors including solar radiation, temperature, day length, and sea ice cover vary predictably along this gradient and are likely to exert a significant influence on ecological processes (Howard-Williams et al., 2006). In these conditions it may be expected that species composition will change and species diversity may fall as latitude increases with the few remaining species adapted to extreme southern conditions. Differences in biomass and community composition within fast ice in two locations separated by five degrees of latitude, the same interval of the present study, along the Victoria Land coast were found by Ryan et al. (2006) and were attributed to significant differences in the physicochemical characteristics of each site.
However, RDA analysis explained about one-third of the total pigment variance, suggesting that pigment assemblages are affected by other variables. It is recognized that the latitudinal gradient approach is complicated by local variability at each latitude due to altitudinal differences, topography, and microclimate (Howard-Williams et al., 2006). Hodgson et al. (2001, 2004b) found that geographical location, water depth, and conductivity can affect the composition of sedimentary pigments. The more saline lakes were characterized by higher abundances of chl b and carotenoids of green algae; diatom carotenoids occurred in all lakes but were relatively less abundant in freshwater ponds (Hodgson et al., 2004b). However, the environmental factors influencing mat composition and structure are complex and differ among the various taxa; also species of the same genus have different distribution. It was suggested that the oscillatorian species were likely distributed based on different conductivities of ponds in the McMurdo Ice Shelf, whereas Nostocales abundance seemed to be independent of conductivity (Jungblut et al., 2005). Likewise, in both continental and maritime Antarctica, Prasiola crispa (Chlorophyta) is found in the vicinity of bird colonies where there is considerable nutrient enrichment, whereas P. calophylla is found in habitats not experiencing nutrient enrichment and is absent from areas where there are high salt concentrations (Broady, 1989). In the Signy Is. lakes, Pearce (2005) found that the bacterioplankton community structure cannot be explained by physicochemical parameters alone, although nutrient concentrations and the timing and duration of ice cover can determine changes in these communities; biotic factors are probably also important. Latitude and altitude are surrogates of spatial environmental variations, and the distribution of compounds is probably a function of many environmental proprieties and factors (e.g., temperature, ice cover persistence, UV radiation, and nutrient availability) acting at different spatial scales.
Lacustrine water productivity varied by one order of magnitude in 15 subarctic lakes distributed within an altitude gradient reflecting a temperature gradient. This variation was mainly caused by variations in the duration of the ice cover and especially in the supply of organic carbon and nutrients (Karlsson et al., 2005). As for other terrestrial environments, Freckman and Virginia (1997) found that a variety of soil factors may together define suitable or inhospitable habitats for soil nematodes at local and regional scale; Powers et al. (1998) in the Taylor Valley soil did not identify a single soil property that fully explains the distribution, biodiversity, or community structure of these invertebrates.
Conclusions
In summary, the results from this study:
were used to compile a detailed database on sedimentary pigments that includes a larger number of compounds than found by previous studies and expands our knowledge of lakes in a relatively unknown but limnologically important region of Antarctica;
show that Cyanophyta is the most abundant and widespread algal group, followed by Chlorophyta and Bacillariophyta, in agreement with microscopy studies conducted by other researchers in a few lakes in the same area;
confirm that the density and diversity of autotrophic communities increase in naturally euthrophized water bodies without an ice cover located in coastal ice-free areas;
show that environmental variables, particularly latitude and altitude, are the main factors affecting pigment distribution;
support the usefulness of a pigment-based approach in the study of lacustrine biotic components.
Acknowledgments
Research was supported by funding from the PNRA (Programma Nazionale di Ricerche in Antartide). We thank anonymous referees for comments on this manuscript.