Interannual meteorological variability and its effects on a lake from maritime Antarctica
- First Online:
- Cite this article as:
- Rochera, C., Justel, A., Fernández-Valiente, E. et al. Polar Biol (2010) 33: 1615. doi:10.1007/s00300-010-0879-8
- 106 Views
The present study shows the occurrence of remarkable interannual variation in the meteorological conditions at Byers Peninsula (Livingston Island, South Shetlands Islands, Antarctica), in which one of the summers was significantly colder than the others. Within this climatic scenario, a limnological study was carried out at Lake Limnopolar during three consecutive summer seasons (2001/2002, 2002/2003 and 2003/2004). The year-to-year meteorological variation observed during this period resulted in marked differences in the timing and duration of the ice-free period. As a result, physical and chemical conditions changed and were followed by variations in the biological characteristics of the lake. More significant dissimilarities took place during summer 2003/2004 relative to the preceding years. This season was characterized by a delay of 55 or 25 days in the ice-out timing compared to 2001/2002 or 2002/2003, respectively, and also a much shorter ice-free period. Higher algal and bacterial abundances in the surface layers occurred at the onset of ice melting due to increases in nutrients and light availability. The trophic interactions could also be affected by ice-out timing, as a consequence of the prolongation of the ice-cover period. From our findings, we describe links between the meteorological variations during those 3 years and the shifts in the water bodies, pointing out their high sensitivity to environmental changes that may occur at different time-scales. Furthermore, our results emphasize how the interannual meteorological variability needs to be investigated as a triggering factor of the limnological variations to understand the effects of global change on limnetic ecosystems in Maritime Antarctica.
KeywordsByers PeninsulaLimnologyMeteorologyIce dynamicsCopepodsMicrobial Plankton community
Lakes, ponds, seepages and streams are typical features of the landscapes found in the Maritime Antarctic region (Ellis-Evans 1996). Some sites, such as Byers Peninsula (Livingston Island, South Shetland Islands), are ice-free during the austral summer and have a high number of aquatic ecosystems (Toro et al. 2007). These ice-free areas are typically smaller than 100 km2. The watersheds are also reduced in size, and the lakes within represent an integration of the different aspects related with the landscape and run-off. Lakes are especially sensitive to the variations taking place in the watershed such as freeze–thaw cycles and snow cover (Quesada et al. 2006), rendering these ecosystems an important baseline for environmental studies (Williamson et al. 2009).
Lake dynamics greatly differ in high latitude Continental and Maritime Antarctica. Permanent ice-cover in the lakes of the Continental region results in small variations in physical and chemical gradients (Spigel and Priscu 1998; Roberts et al. 2000). In addition, the pelagic production in perennial ice-covered lakes is mainly based on the internal carbon budget (Dore and Priscu 2001), and interannual variation in phytoplankton dynamics is suggested to respond mainly to wintering over strategies and trophic interactions between organisms (McKnight et al. 2000). The situation is completely different in the Maritime and in other lower-latitude Antarctic regions, as the warmer climate regime allows ice melting in summer, when lakes mix completely. The ice melting results in allochthonous inputs of nutrients that subsidize lake productivity (Priscu et al. 1998).
In addition, phytoplankton dynamics is also different in lakes in Continental and in Maritime Antarctica. While Continental lakes seem to lack patterns of species succession during the summer (Spaulding et al. 1994; Lizotte et al. 1996), lakes in Maritime Antarctica show important changes in species composition over short periods, although they do not fit with a conventional scheme of community succession (Ellis-Evans 1996). The ecological succession found in these ecosystems might be related to local environmental changes, as the ice-cover dynamics could hypothetically affect the trophic interactions and patterns of biomass transfer to higher trophic levels (Camacho 2006).
The coupling between phytoplankton species succession and changes in the stability of the water column, which mainly depends on ice-cover dynamics, has been described for some Antarctic lakes (Priddle et al. 1986; Izaguirre et al. 1993). For instance, under low turbulence conditions (under ice-covered water columns), only species with reduced sinking rates are able to remain in the water column. The ice-free period allows an increase of light availability through the water column and a significant nutrient input via run-off, which may induce important changes in lake functioning such as nutrient fluxes (Blenckner 2005) and primary production (Adrian et al. 1999; Park et al. 2004; Mckenna et al. 2006). In particular, the underwater light regime in polar lakes plays a major role in the functioning of lake ecosystems (Tanabe et al. 2008).
Interannual climatic variations in the Antarctic Peninsula region have been found to be higher than in other Antarctic regions, and this is probably related to the sea-ice extent (King 1994). In addition to this variability, we hypothesize that terrestrial and limnetic ecosystems in this region are very sensitive to small changes in air temperature because the region has milder summer temperatures that are very close to the freezing point. In that way, small variations in air temperature can promote important changes in the ice-cover and thus in the lake dynamics during the spring–summer period, in which melting occurs. Additionally, winter snow accumulation may differ from year to year, and this could also play a role in the summer melting dynamics of both the lake ice cap and the snow cover within the catchment. The latter greatly influences lake hydrology, and, consequently, attention must also be paid to the winter climate in order to understand the summer lake dynamics. To test this hypothesis, some relevant limnological parameters, physical, chemical and biological in nature, were obtained during three consecutive Antarctic summers in which the climatic conditions and, hence, the ice-cover dynamics were very different. Both survey periods of summers 2001/2002 and 2002/2003 represented mainly open-water phases, whereas the main part of the survey in the summer of 2003/2004 was conducted when the lake was still covered by ice.
Materials and methods
The climate at Livingston Island is maritime and less extreme than in Continental Antarctica, with mean summer temperatures in the range of 1–3°C, daily maxima up to 10°C and daily minima down to −10°C. In winter, minimum temperatures can reach −27°C, and winter maxima are always below 0°C. Precipitation is much higher than in most of Continental Antarctica, with annual mean values of 700–1,000 mm (Bañón 2001; Van Lipzig et al. 2005). The region is snow covered for at least 7–8 months a year, and snow packs can persist throughout summer in some accumulation areas. Wind directions are variable although westerlies or easterlies dominate.
Lake Limnopolar (unofficial name but used in scientific literature and congresses) is a small lake (estimated area 22,172 m2), with a maximum depth of 5 m. It is located c.a. 90 m a.s.l. on the southwestern part of the central plateau of Byers Peninsula (Fig. 1), and it is an open system with a small stream as outlet. The lake has a scarcely vegetated drainage watershed of 0.581 km2 and most of the catchment run-off flows into the lake through a small stream. The bottom of Lake Limnopolar is covered by the aquatic moss Drepanocladus longifolius. There is another small drainage lake within the Lake Limnopolar watershed, Lake Somero (unofficial name), which is located 200 m upstream. Its maximum depth is 0.7 m, and its outlet flows and mixes with the inlet water of Lake Limnopolar.
Meteorological data and freeze date estimation
The meteorological characteristics at Juan Carlos I Base were measured with an automatic meteorological station (AMS) equipped with a SEAC (until 2005) and a Geonica (since 2005) datalogger and using temperature and humidity sensors at 1.5 m above the ground, with a wind speed and direction at 10 m. Data were stored every 10 min. The extended summer temperature data (November to February, both included) from this station have been measured since the summer of 1987/1988.
Meteorological data in the lake site were obtained from an AMS equipped with a Campbell CR10X logging unit. The AMS was located between Lakes Limnopolar and Somero (Fig. 1), at 2 km from the coast. It provided continuous data, with exceptions during some midwinter periods of battery malfunctions. This station registered the typical meteorological variables as well as the water temperature at the bottom of Lake Somero. The recorded data at intervals of 30 min consisted of the mean value of 60 measurements (temperature) or the integration of 1,800 measurements (global radiation). Precipitation was measured daily during three summer periods (2001–2004) by using a simple rain gauge placed about 2 km south of the lakes.
Lake ice-off dates, defined as the first day on which the main basin of the lake was totally ice free, were determined by visual observations. Variations in the thicknesses of snow and ice and the levels of water were also monitored during the summer period. Snow and ice thicknesses were measured with a graduated stick, and water level with a portable depth echosounder (Plastimo Echotest II). From the end of February, logistical constraints prevented further studies. For this reason, we followed the modeling ideas of Cahill et al. (2005) to estimate freeze dates, defined as the first day with a covering of ice on the main basin that did not thaw for at least 3 days. The freezing model was based on the assumption of different water temperature fluctuations before and after the development of ice cover, measured in our case by the standard deviation of the five subsequent mean daily temperatures. When the lake was ice-covered, the deviations fell and stabilized near zero. Therefore, we estimated the freeze date by locating the breakpoint in the 5-day standard deviation series pattern between two thaw dates. The breakpoint corresponds to the day that the sum of squared residuals was a minimum, assuming different linear models at both sides of the breakpoint. The freeze date search was limited to the days with mean water temperatures ≤0°C. The continuous temperature record for Lake Limnopolar was only available for 2002. Nevertheless, temperature data for Lake Somero (which has similar freezing-thawing characteristics) were available for the 3 years (2002, 2003 and 2004). Thus, the ice-cover duration was estimated for Lake Somero and assumed to be similar to Lake Limnopolar, although visual observations were only possible for the thawing period.
In situ measurement of physical and chemical parameters
Field work was carried out at Lake Limnopolar over three consecutive summers from December 2001 to February 2004. The routine sampling site was located at the point of maximum depth. Vertical profiles of temperature were obtained with a 6920 YSI multiparametric probe every 0.2 m. During the period when the lake was covered by ice, sampling devices were lowered through a hole in the ice drilled with a motorized ice auger. When the ice melted out, sampling was carried out from a boat at the same point. The profiles of broadband photosynthetically active radiation (PAR; 400–700 nm) were measured as quanta (μmol m−2 s−1) at 0.5-m intervals with a 2π sensor model Li-192SA, with uniform response in the range of 400–700 nm (PAR) attached to a LI-COR datalogger, model Li-1000. When the lake was ice-covered, these profiles were measured by lowering the sensor from beneath the ice. From the obtained data, extinction coefficients (kPAR) for the complete ice cap and through the water column were calculated by regression in accordance with the log-transformed equation of the Lambert–Beer law.
A series of thermistors was suspended from an anchored buoy in the summer 2003/2004 to record temperatures at short time term intervals. It was made up of 5 thermistors (Onset Tibdit) placed equidistantly down the water column, fixing the position of the uppermost relative to the ice cap. They registered a measurement each 30 min from December 29th until February 12th. For all calculations, the data were interpolated to acquire 0.25-m intervals using a spline function (Stoer and Bulirsch 2002).
Water column stability calculations
The stability of the water column was estimated numerically by calculating the profiles of Brunt–Väisäla frequency (N2) as described by Lemmin (1978). N2 represents the stability coefficient expressed in s−1, based on the density gradient against depth. For density calculations, values were obtained from the temperature and conductivity dates acquired in vertical profiles. For mathematical details of the calculations, see Fofonoff and Millard (1983). From the results obtained, we assume that the pycnocline occurs where N2 is maximal.
Nutrients and biological sampling
Sampling for biological and chemical analysis was performed in parallel to the physical–chemical survey. Water samples were taken from surface layers of the lake (between 1–2 m) using a Kemmerer bottle. These were obtained at different intervals during different summers as follows: in 2001/2002, three times from mid-December to February; in 2002/2003, only once in January; and in 2003/2004 with a weekly frequency from the third week of December to the second week of February. Sub-samples for the analysis of major dissolved nutrients (combined nitrogen forms and reactive phosphorus) were filtered in situ through pre-combusted glass fiber filters (Whatman GF/F) and were collected in acid-washed polyethylene bottles. Subsequently, they were frozen until being processed in the laboratory following the analytical recommendations of RiSCC (Huiskes and Quesada 2002), using standard analytical methods (APHA 1992). Ammonia (NH4+) was analyzed by the phenol-hypochlorite method. Nitrate plus nitrite (NOx) was determined as nitrite following the Griess method, after the reduction of nitrate to nitrite by passing the samples through a reductive cadmium column. Soluble reactive phosphorus (SRP) concentrations were determined using ascorbic acid reduction of the phosphomolybdate complex following the Murphy and Riley method. Total nitrogen and phosphorus concentrations were obtained following the same method described, respectively, for NOx and SRP after performing alkaline and acid digestions, respectively, on unfiltered samples.
Samples for determination of the abundance of heterotrophic bacteria (HPP) and nanoflagellates (NF) were fixed with a buffered formalin solution (2% v/v final concentration). Bacterial counts were done by epifluorescence on a Zeiss-III phase contrast microscope on DAPI (4′,6-diamidino-2-phenylindole)-stained samples concentrated (3–5 ml) on black polycarbonate filters (0.2 μm). During filtration, a 0.8-μm-pore cellulose acetate backing filter was used to obtain a uniform distribution of cells. Nanoflagellates were also stained with DAPI and were counted similarly to bacteria. In this case, preparations were obtained by filtering a volume of up to 30 ml of sample with low vacuum (<15 in. of mercury) onto a 0.8 μm Isopore GTBP (Millipore). Samples for determination of the abundance of ciliate protozoa were preserved in glass bottles with Lugol’s iodine solution (1.5% final concentration) for microscopic examination. In the laboratory, their abundances were determined with an inverted microscope at 1,000× using the Utermöhl (1958) sedimentation method. Samples for counting zooplankton were preserved by adding a formalin solution to a final concentration of 4% (v/v). Counts were performed under a Nikon® binocular microscope, model TMS.
Pigment (chlorophyll-a and taxa-specific carotenoids) analyses were performed by HPLC as described by Vincent et al. (1993). Samples were filtered at the field site through glass fiber filters (Whatman GF/F) and were frozen immediately. The peak identities were determined by comparing retention times and spectra with pure standards purchased from DHI (International Agency for 14C Determination and Pigment Standards). The amount of pigment was quantified against the curves obtained with the standards by integration of the area under the peak at 440 nm. The chlorophyll-a (Chl-a) concentration was used to quantify the abundance of phytoplankton and its changes were interpreted as changes in algal abundance. Fucoxanthin and lutein were determined as taxa-specific carotenoids for the quantification of the relative abundances of chrysophytes and diatoms using the former and of chlorophytes and prasinophytes using the latter.
Environmental conditions and ice cap variations
Meteorological conditions in Byers Peninsula during the summers of 2001/2002, 2002/2003 and 2003/2004
Air temp. (ºC)
Soil temp. (ºC)
Water temp. (ºC)
Radiation (Kj m−2)
Wind speed (km h−1)
Max wind speed (km h−1)
The different meteorological conditions gave rise to clear differences in the durations of the ice-free periods. The ice-free days in Lake Somero varied from 96 ice-free days in the summer period of 2001/2002 (from December 22nd 2001, to March 27th 2002) to 41 in the summer period of 2003/2004 (from February 11th 2004 to March 23rd 2004) (Fig. 2). This means that the ice-free period in the summer of 2001/2002 was more than twofold longer than that in the summer of 2003/2004. Visual observations and modeling efforts indicated that the ice dynamics in both lakes (Lake Somero and Lake Limnopolar) were very similar.
Optical characteristics of the water column
Thermal condition of lake and water column stability
The thermal structure of the water column in 2003/2004 under the ice-cover is shown in Fig. 6. At the beginning of our measurements, the water column temperature was quite homogenous, being somewhat warmer at the bottom (1°C). Later on, an evident inverse stratification built up, in which the colder water (close to 0°C) was at the surface just under the ice, whereas the deepest layers remained near 4°C (Fig. 6). During this time, and mainly when the ice was thinner and more irradiance was available, circulation increased in the lake, with convective plumes migrating gradually through the water column and transporting heat downwards. This resulted in a progressive warming of the deep water layers from mid-January to the first week of February, with an intense temperature gradient of 3.5°C from the surface to the bottom. This thermal discontinuity progressively decreased until the complete ice-out, which triggered the water column homogenization. From that moment, the thermal regime of the water column followed the diurnal temperature variations.
Similarly to temperature, conductivity profiles in the summers of 2001/2002 and 2002/2003 were characterized by totally uniform distributions. Variations during the season were also narrow; while in 2001/2002 the mean values through the water column increased gradually from 52 to 77 μS cm−1 in the summer of 2002/2003, the conductivity was higher in mid-January (76 μS cm−1) than in mid-February (60 μS cm−1). When the ice cap was present in the summer of 2003/2004, conductivity was higher close to the bottom than in the rest of the water column (137.9 μS cm−1 at the bottom and an average of 53 μS cm−1 for the whole water column, excluding the bottom at the end of December). From mid-January, conductivity gradient was more pronounced from around 2 m to the bottom, in parallel to the temperature gradient, indicating the ice melting on the surface. Later on, when the ice had completely melted in February, the conductivity became constant with depth.
Soluble reactive phosphorus (SRP) concentrations in different years varied from undetectable levels (<0.03 μM) to around 0.1 μM (Fig. 8). As a general trend during the summer seasons of 2001/2002 and 2003/2004, the concentrations decreased as the seasons advanced, with higher concentrations occurring during ice-cover periods. After the ice thaw, the SRP concentrations remained similarly low.
Trends in the particulate nitrogen (PN) and phosphorus (PP) also substantially diverged between the different summer periods. The PN concentration was stable (ranging between 7.2 and 9.7 μM) in 2001/2002, however, but a fourfold increase in the period close to the ice thaw followed by a marked decrease was observed in 2003/2004. The only value available for 2002/2003 was in the range of values found for the other two seasons. The PP concentration followed similar trends as PN in 2003/2004; however, it decreased across the summer in 2001/2002. Again, the ice conditions showed an important influence on the PP and PN, with higher values before the ice thaw. The TN/TP ratio, estimated as the ratio between the DIN+PN/SRP+PP, varied broadly between 22.2 and 94.3 in 2001/2002 after the ice thaw, although lower values were recorded in periods prior to the ice melting (Fig. 8). The only available datum for 2002/2003 also showed a high ratio after the ice melting. This fact could represent the variable limitation of N and P related with the ice conditions.
Changes in the taxon-specific carotenoids normalized to Chl-a are shown in Fig. 9. The fucoxanthin/Chl-a ratio (as an indicator of chrysophytes and diatoms abundance) showed a constant increase along time, reaching relative concentrations of 0.23 and 0.1 μg (μg Chl-a)−1 in the sampling periods of the summers of 2001/2002 and 2003/2004, respectively. The ratio of the only datum available in 2002/2003 was in the range of the other two sampling seasons.
Lutein concentrations in the sampling period of the summer of 2001/2002 (as an indicator of green algae and prasinophytes) followed a similar pattern to that of fucoxanthin. The lutein/Chl-a ratio showed a fourfold increase when the lake was ice-free compared to the first sampling date, when the lake surface was still partially frozen. However, no significant increase was observed for the lutein/Chl-a ratio in 2003/2004.
Prokaryotic and protistan plankton
The surface densities of different microbial planktonic populations during the studied years are shown in Fig. 9. The heterotrophic picoplankton (HPP) abundances varied in a narrow range during the different seasons, although they showed different trends depending on the ice conditions. Thus, during the sampling period in the summer of 2001/2002, high abundances up to 2 × 106 cel mL−1 were noticed when the ice was melting out. Then, under ice-free conditions, bacterial densities fell to levels between 1.3–1.5 × 106 cel mL−1, which nearly coincided with the available data from 2002/2003 in the same period. In the summer of 2003/2004, bacterioplankton (HPP) densities showed a progressive increase from 0.8–1.6 × 106 cel mL−1 that peaked when the lake was totally ice-free. These values were comparable to densities observed during 2001/2002 under the same circumstances.
Nanoflagellate (NF) populations were composed by both heterotrophic and autotrophic forms, the later being relatively more abundant during the summer of 2003/2004 compared to previous years (data not shown). Protists of 3–5 μm long dominated the heterotrophic subset, which were assigned to chrysomonads (likely Spumella and Oikomonas). Another fraction of the community belonged to unidentified small plastidic flagellates that could tentatively be classified as chrysophytes (Ochromonas-like and Chromulina sp.); however, others could be prasinophytes species. Moreover, the more recognized pigmented forms were Chrysophytes composed mainly by two forms assigned to the genus Pseudokephyrion, and these were especially abundant. The variation pattern of NF was coincident with the Fucoxanthin/Chl-a ratio. Total numbers at the surface waters varied similarly in the summers of 2001/2002 and 2003/2004, ranging from 115–526 cel mL−1 and 131–457 cel mL−1, respectively, with higher numbers occurring during ice free and melting phases, respectively. In 2002/2003, the abundance was below 200 cel mL−1.
Ciliated protozoa diversity was particularly low in Lake Limnopolar; hence, during all 3 years, the assemblage was mainly composed by few nanoplanktivorous species of prostomatids in which Balanion planctonicum clearly dominated. The ciliates community appears to increase after the ice melting. In the summer of 2001/2002, ciliates reached high densities between 2.2 × 103 and 2.8 × 103 ind L−1. The density observed on February 15th, 2003, when the lake was experiencing similar conditions, was also of this order. In contrast, in 2003/2004, the abundances ranged over low levels between 0.2 × 103 to 1.2 × 103 ciliates L−1.
In all the studied years, the only relevant metazooplankton species has been the copepod Boeckella poppei. Some individuals of the cladoceran Macrothrix oviformis (after Kotov 2007) have also been found (Toro et al. 2007, sub M. ciliata), but were clearly associated to the bottom as a benthic species. Rotifers have also been present, mainly represented by specimens of the genus Notholca, although they were rare in samples and their numbers never exceeded densities of 0.2 ind L−1.
Regarding copepods, the densities in surface waters (Fig. 9) differed between both summers of 2001/2002 and 2003/2004. Thus, during 2001/2002, the densities detected at the surface of the lake were always below 1 ind L−1, whereas in 2003/2004 the under-ice abundances were in a higher range between 1–5 ind L−1. However, these data on abundance correspond to surface waters, whereas copepods are always more abundant in deep waters. As our more precise data about the copepod distribution in the water column demonstrated that this species is able to perform vertical migrations in the lake, these data on abundance might not necessarily be interpreted as significant interannual differences. On the other hand, the age structure of the population of Boeckella poppei changed markedly among the different years when precise dates were compared (Rochera et al., in preparation). As the relative proportion of the different age stages differed at certain dates of the summer periods, with a delay in the dominance of adult forms when ice melting was delayed (such as in the 2003/2004 summer), this can be used as a good descriptive factor of the meteorological effects on the biota.
Studies on the variation of ecosystem functioning driven by meteorological changes are of great interest in building global change models. In this work, we present data on meteorologically driven ecosystem variation over a short period of time (3 years) representing the wide natural variation of these extremely sensitive non-marine aquatic ecosystems as a useful tool for future climate change studies in one of the regions on Earth where warming is more pronounced (Quayle et al. 2002).
The studied area presents mild summers with average temperatures slightly above 0°C (Fig. 2a). The proximity of summer air temperatures to the freezing point makes the aquatic ecosystems of Livingston Island very vulnerable to slight changes in temperature, which can trigger remarkably longer or shorter ice-free periods because the ice duration in a lake is 60–70% a function of the air temperature (Palecki and Barry 1986; Livingstone 1997). This effect is clearly illustrated in Figs. 2 and 3b, in the summer of 2002/2003, an earlier decrease in temperatures in February triggered the freezing of the lake water a month earlier than is typical for the area. However, some weeks later, the temperature increased again and promoted a new thaw.
The meteorological variation observed in the three consecutive years indicated that the summer of 2003/2004 was much colder than the previous years, as temperature data are available (1987). This variation notably influenced the length of the ice-free period in the three consecutive years investigated, producing a difference of 47% in the duration of the ice-free period in the 3 years. The estimated difference of 55 days in the ice duration could be very relevant in terms of light availability, temperature and stratification, shifting the ecological relationships. In temperate lakes in Canada, the ice-off dates in a 6-year series ranged 34 days (Wynne et al. 1996). The ice-cover duration in Lake Müggelsee (Germany) varied over 100 days in consecutive years (Adrian et al. 1999). This variation in the duration of ice-cover has been suggested as a potential uncoupler between the trophic elements (Winder and Schindler 2004). This is also likely to modify most functional aspects in the relationship between the lake and its catchment (Camacho 2006), as the productive period would be longer both in the lake (represented by phytoplankton but also by aquatic mosses) as well as in the catchment where active microbial mats cover wide areas and can act as a nutrient source for the lake (Fernández-Valiente et al. 2007).
Lake Limnopolar shows a thermal pattern between cold and temperate thereimictic lakes, which is a regular condition of lakes present at the site (Toro et al. 2007). This same mixis pattern has been reported for other sites from the Maritime Antarctic region including Deception Island (Drago 1989), the South Orkney Islands (Heywood 1967) and King George Island (Drago 1980). It is characterized by circulation only during the summer season. Lakes in this case present a seasonal cycle similar to cold monomictic, although without summer stratification. A progressive increase in conductivity during the summer seems to be a common trend in polar lakes (Borghini et al. 2008 and articles cited therein) and may respond either to the increase of drainage during melting processes or to evaporation processes. In ice-free areas such as Byers Peninsula, interactions between the aquatic ecosystem and the surrounding land are greatly restricted during the winter by the presence of ice, but during the thaw season they become more intense, just when biological activity increases.
Ice-cover is also very important in terms of the stability of the water column, as the wind effect is reduced and inverse water stratification remains under the ice cap. In this way, data obtained in the summer of 2003/2004 provide a suitable observation of seasonal patterns of thermodynamic and mixing processes in an Antarctic lake throughout the course of the thaw period. It is evident that the presence of ice-cover prevents both evaporation and wind effects, so it allows for storing of the heat entering into the lake, which finally results in the observable progressive increase of water temperatures. In 2003/2004, from January 24th to the beginning of February, the lake water warmed up apparently by an increase of the incident solar radiation to the surface due to snow and ice loss. It is known that the heat fluxes provoked by solar radiation gain relevance, coinciding with the thickness reduction of snow cover (Bengtsson and Svensson 1996), and interestingly just before the break-up of ice-cover (Bengtsson 1996), as in our case. During these periods, the incoming radiation may warm the water immediately under the ice, producing convective plumes that penetrate down into the water column beneath (Vincent 1988). The variation of weather conditions appears to be also be the cause of acceleration in the ice-off.
Based on nutrient and chlorophyll-a levels, Lake Limnopolar may be considered as ultra-oligotrophic, in which the carbon transfer pathways based on microbial populations may play a key role. Ice melting seems to be an important period in the lake biological dynamics, as bacterial numbers as well as those of the primary producers are higher during this period. Additionally, we hypothesize that the DOC released by the pelagic primary production may play a secondary role as fuel for bacterial production, given the low Chl-a pelagic concentration, and other sources such as the benthic production (based on the aquatic mosses carpeting the lake bottom) or the watershed dissolved organic carbon (DOC) released by microbial mats and other autotrophic components may sustain the heterotrophic production in the lake (Camacho 2006).
In polar summers, irradiance availability for aquatic organisms experiences high variability due to the ice and snow cover. In Lake Limnopolar, the underwater light regime follows a progressive change toward higher transparency. Although benthic organisms in permanently ice-covered polar lakes are adapted to low irradiance (Walton and Doake 1987; Hawes and Schwartz 2000), their primary production can be quite limited in moderately high latitudes (as in Byers Peninsula) because the number of light hours is limited in spring, the solar angle is low and the ice and snow absorption coefficients are very high (Fig. 4). Snow on the ice-cover and also the white ice limit the light penetration into a lake more than black ice (Petrov et al. 2005). Therefore, local meteorology in terms of snowfall events can play an important role in the energy exchange processes during the ice-cover and melting periods. Changes in the ice thickness, snow cover and ice cap duration will represent important variations in the light availability for the underwater community, potentially influencing both pelagic and benthic primary production and thus the bacterial production in the whole ecosystem.
Our results in Lake Limnopolar indicate that before complete ice melting takes place, the phytoplankton community blooms (which can be associated to nutrient pulses related to ice melting) with a synergistic effect with the increase in light availability. Similar algal blooms have also been described in Otero Lake, located in the Antarctic Peninsula (Mataloni et al. 2000). There, authors proposed high hydrologic residence times as the reason behind this high algal abundance, as washout is impeded. Otero Lake is, however, a eutrophic lake, whereas Lake Limnopolar has a much lower trophic status. In such an oligotrophic lake, the increased nutrient availability, due to ice thawing, and the increased light penetration because of the already thinned ice cap could likely explain the observed algal bloom. Hence, nutrient concentrations can increase at surface layers as a consequence of debris-containing ice melting because, as observed in other Antarctic sites (Roberts and Priscu 2004), these nutrients might be accumulated in the ice due to the particle movements promoted by wind blowing on exposed soils within the catchment or by aerosols load (Prendez et al. 2009). In the case of Lake Limnopolar, this could even be stronger due to the occurrence of snow covering the ice cap, which also contains nutrients originating from the atmospheric deposition (mainly nitrogen). The increased nutrient availability coincides with the onset of light availability due to the decay of ice-cover. Certainly, during the summer of 2003/2004, the thermal convection in the lake, which promotes nutrient movements, started at the beginning of January. However, the ice conditions at this time limited the light availability, and the requirements for algal growth were not totally met, so a bloom did not occur. With our sampling resolution, we did not detect the suggested increase of inorganic nutrient concentrations likely occurring immediately below the ice when melting; however, the extremely low concentrations found in this lake indicate a strong nutrient limitation and thus a very fast uptake could have taken place.
The variation in the moment of ice thawing could be extremely important for the biological activity in the lake because of the impairment between different life cycles of the organisms (e.g. copepods vs. nanoflagellates). Also the duration of the ice-cover could represent important effects on the ecological dynamics of the lake, allowing longer or shorter periods of production in the lake and, secondarily, in the catchment, with both of these influencing the nutrient supply to the different components of the food web.
The evolution of fucoxanthin concentrations followed the Chl-a concentrations in the 2003/2004 summer, even during the algal bloom. This carotenoid is held both by chrysophytes and diatoms, but our microscopic counts seem to indicate that chrysophytes were mostly responsible for this algal growth. In the summer of 2001/2002, during ice-off conditions, fucoxanthin increased in parallel with nanoflagellate abundance. However, in 2003/2004, the evolution of the pigment ratio did not exactly follow the nanoflagellate counts. This could be due to a major role of the heterotrophic forms, but it also might be related to the cells’ adaptation to the bright light experienced after the melting of ice and changes in the cellular quota of these pigments (Tanabe et al. 2008). The development of chlorophytes, deduced from the ratio of lutein/Chl-a, might in part be explained by the large size cells which might be related to the turbulence periods of the lake after the ice melt, although prasinophytes (pico- and nano-sized), which also hold this pigment, are other possible important components of the lacustrine phytoplankton likely contributing to these variations.
Trophic interactions taking place in the lake might also be affected by the ice-cover period. Our data indicated a delayed maturation of copepods in the summer of 2003/2004 compared to previous years, due to either a delay in hatching events or a diapause during some copepoid state under the longer ice-cover persistence. These variations may not match with the timing of the algal bloom prior to the ice melt, with dramatic consequences for the copepods population, as it could be the only opportunity to obtain a high quantity and quality of food for herbivorous zooplankton populations. Seebens et al. (2009) in a long-term study on the variations of copepods (Cyclops vicinus) related to variable phytoplankton bloom phenology showed that a delay in maturation could be a strategy to guarantee the match between copepods and higher food availability. Nevertheless, higher survival rates in copepods were dependent on phytoplankton bloom timing, and these were higher in years with earlier phytoplankton blooms. It is unknown if these strategies will be flexible enough to cope with future climate warming. In any case, manipulative experiments (Camacho et al., in preparation) have shown that copepods from Lake Limnopolar can also use other food sources other than phytoplankton, although even more complex interactions among the food web members could also be determined by both interannual meteorological variability and the trend of climate warming. As an example, other trophic processes could also be affected by variations in ice dynamics. For instance, the bacterial carbon pool constitutes an important source of food for both ciliates and heterotrophic nanoflagellates as is illustrated by the decline in bacterial numbers when the bacterivorous organisms peaked (Fig. 9).
In this work, we illustrate how the meteorological conditions were quite different in three consecutive years and we explain how these conditions governed the duration of the ice-free period in the studied lake. The variation in the duration of the ice-free period directly affected the physical and chemical characteristics of the water column, which in turn influences the structure of the biological community. For instance, taking into consideration the main role of the microbial loop in these lakes, and given that ice dynamics might affect the relative contribution of internal and external carbon sources, it is possible to state that depending on the ice conditions in lake, internal primary production or water catchment production may shift as the main sustaining source for the food web of the lake.
In conclusion, our findings show how several limnological aspects, both physical and biological, could be affected by environmental forces, especially the ice-cover period during the productive season. Our study provides a demonstration on how ice-cover dynamics was especially sensitive and was thus subjected to variations. In our opinion, further attention has to be paid to the fact that these year-to-year variations may be large and thus may complicate the explanation of long-term variations attributed to climate change unless we better understand these dynamics. Consequently, to more precisely develop the long-term trends in lake ecosystems, studies would have to be large enough to integrate the interannual variability that we found in the present study.
For a better understanding of the responses of lake ecosystems to climatic variations, and given that Byers Peninsula comprises one of the richest limnetic areas of the Maritime Antarctic region, this area has been designated as an International Site of Reference for Coastal, Terrestrial and Limnetic studies (Quesada et al. 2009). It could therefore be appropriate to initiate long-term limnological research at this site in the same manner as that conducted in other regions of the continent (i.e. the LTER program in Dry Valleys). This is a general requirement of ecological science, as long-term ecological research appears to be the main, and perhaps the sole, way to improve the calibration of the models with respect to the adaptive responses of biological communities and ecosystems to the current environmental changes occurring on the Earth.
We are grateful to Ministerio of Educación e Innovación (Spain) that funded this research by the projects REN2000-0435-ANT to Antonio Quesada, CGL2005-06549-C02-01/ANT to Antonio Quesada, CGL2005-06549-C02-02/ANT to Antonio Camacho (in this case with European FEDER funds) and CGL2007-29841-E to Antonio Camacho. We thank our scientific colleagues and field assistants for their help in the field and for fruitful scientific discussion. We also very much appreciate the logistic help and support from the UTM (Maritime Technology Unit, CSIC) and from the Las Palmas crew (Spanish Navy) that made this expedition possible. We are grateful for the comments and suggestions of three anonymous reviewers.