This thesis studies the microbial communities inhabiting freshwater ecosystems of Byers Peninsula (Livingston Island, Antarctica). This is an ice-free area of around 60 km2 that holds numerous lakes, ponds, rivers and wetlands. As a deglaciated region, the functioning of aquatic ecosystems is very closely linked with the surrounding land. So, when snow melting occurs during summer, interactions with the catchment become more intense, and coincide with enhanced biological activity. We conducted a systematic survey in the most representative lakes during some consecutive austral summers between 2002 and 2006. Most of these lakes are located on a plateau, where the landform facilitates water retention and they show well-defined surface outlets. According to their nutrient concentrations and the phytoplankton biomass, these lakes range merely from ultra-oligotrophic to oligotrophic. Our surveys have revealed the existence of a large standing stock of mosses (Drepanocladus longifolius) at the bottom of some of them, thus suggesting that primary production could mainly be allocated in the benthic compartment. On the other hand, the shallow lagoons located in coastal areas usually show bigger catchments, which are largely covered by mosses cushions and plants. These coastal lagoons have a higher nutrient content due to the inputs of elephant seals (Mirounga leonina) dwelling in their vicinities, and somewhat greater salinity due to more exposure to sea spray. Bacterial abundances vary broadly among lakes (0.5 to 6.5 x 106 cell mL-1), thus also underlying this inland-coastal gradient of productivity. Still, these bacterial abundances are greater than is expected for these nutrient-poor lakes. We also performed a multivariate analysis that reveals other factors such as the lakes’ morphometry to explain the more subtle differences in their trophic status. In general, all the lakes hold a few species and apparently exhibit a simple food web dominated by microbial communities. The only relevant metazoan species we found are the copepod Boeckella poppei and the fairy shrimp Branchinecta gaini. Besides, the former is the only species to present a significant role in the pelagic part. Our surveys describe microbial communities dominated by small flagellates species (both colourless and plastidic forms), amoebae and a few ciliates species. These last species are mainly composed of the euplanktonic ciliate Balanion planctonicum, with densities in Lake Limnopolar of up to 2.8x103 ind L-1, just after the ice cover retreat. Otherwise, phytoplankton is composed mainly of diatoms, chrysophytes, picocyanobacteria and chlorophytes. However, these assemblages may be incidental since some species originate from the catchment or are resuspended from the benthos when turbulence in the lake increases. One of our aims was to test the basic principles of the functional ecology in these simple environments. These lakes support plankton in truncated pelagic food webs, where the copepod B.poppei is the capstone predator. A strong trophic connection exists, however, between copepods and the microbial loop. We present a conceptual model of the ecological functioning of one of these lakes, Lake Limnopolar, located on the plateau, which we have adopted as a model study lake. By means of manipulative experiments, we demonstrate the potential existence of a trophic cascade in its pelagic food web. It is mediated by a strong top-down regulation of protozoa populations by copepods, which indirectly benefits pico-sized organisms (both autotrophic and heterotrophic). The experiments showed that these trophic pathways are mainly channelized through ciliates, whose abundance declined markedly in response to increased copepod densities. The selective grazing of protists over picoplankters furthermore favours a shift in the bacterial size structure. Our observations also advocate the incorporation into this food web modelling of an effective recycling of nutrients driven by these copepods. In this sense, we conjecture that a translocation of nutrients from lake’s bottom to surface waters could be mediated by the copepods’ diel vertical migration. The analysis of the isotopic carbon fractionation occurring in Lake Limnopolar indicated that B.poppei profits mainly from pelagic resources. Besides, the liquid chromatography (HPLC) analysis of taxa-specific carotenoids revealed the ability of this copepod to produce a redistribution of the relative abundances of pelagic algal groups because of a differential grazing pressure. For instance, this indicates that chlorophytes (e.g. Ankistrodesmus antarcticus) , but not diatoms and chrysophytes, are favoured when copepods abound. In some periods however, we found that copepods mainly retreated to deep layers, suggesting nektobenthic behaviour. Here, we agree with those observations indicating a certain phenotypic plasticity of this species, in such a way that its feeding mode may vary depending on environmental conditions. In contrast, isotopic fractionation clearly revealed that Branchinecta gaini exploits benthic resources, thus demonstrating a niche segregation between the two dominant metazoans in the lake. We demonstrate that ice dynamics is very much subject to variations due to the year-to-year meteorological variability of the region. We observed, for instance, inter-summer differences in the ice-out timing of 55 or 25 days. We demonstrate that this local variability could be an important factor to explain the limnological dynamics. Therefore, further attention has to be paid to the fact that these year-to-year variations may be large and may thus complicate the explanation of long-term variations attributed to climate change, unless we better understand such dynamics. We partially studied the gains and losses of heat in the lake, and observed differences either with or without a frozen cover. During the ice-free periods, if solar radiation is low, the lake loses sensible heat by the shear produced by wind, although this occurs with some delay. On the other hand, density-driven currents dominate when the lake is to be ice-covered. Seasonal changes in the phytoplankton community structure occur in Lake Limnopolar in relation to ice dynamics. When the lake is ice-covered, light availability is greatly limited and nutrients diffusion is restricted because of the high water column stability. During this period, autotrophic picoplankters, both picocyanobacteria and picoprasinophytes, which usually play a minor role, account for up to around 50% of the total phytoplankton biomass at sub-surface layers. By contrast, greater abundances of larger phytoplankton (chrysophytes and diatoms) and bacteria take place with the onset of ice melting and coincide with increases in nutrients, turbulence and light availability. In other respects, and along the trophic gradient previously defined among the lakes, the relative role of autotrophic picoplankters increases as nutrient content decreases. In general, pelagic primary production is very low in the lakes from Byers; however, the plankton’s heterotrophic component is relatively abundant for such low productive rates. In relation to this, our studies suggest that the benthic communities which flourish in the surrounding areas of the lakes (principally microbial mats) could be a source of allochthonous inputs of organic carbon, as the isotopic fractionation also suggests, thus contributing to fuel bacterial production. This organic carbon may also originate from the benthic mosses growing within the lake. In this sense, an in-depth study into this subject is necessary because, if bacteria are mainly subsidised by allochthonous carbon, the idea that low-productive systems tend to be net heterotrophic is reinforced. The dissolved organic carbon (DOC) concentrations in the oligotrophic lakes of Byers Peninsula were around 1 mg L-1, which seems high enough to sustain bacterial growth. The results indicate that bacteria may compete with phytoplankton for mineral nutrients, which hypothetically may occur if nutrients availability is notably scarce. We also examined the taxonomic and physiological diversity of the benthic microbial communities at the site. For this purpose, we conducted a multi-approach study of three representative microbial mats at the site. Two of the mats (soil and pond mats) were located on the central plateau, whereas the third one (stream mat) was located in the coastal area of the South Beaches on the edge of streams. The microscopic and pigment analyses revealed that the stream mat was dominated by diatoms (genera Navicula, Fragilaria, Stauroneis, Nitzschia, Gomphonema and Pinnularia), whereas the mats’ characteristics of the plateau (soil and pond mats) were dominated by cyanobacteria (Leptolyngbya sp., Oscillatoria spp., Phormidium spp., Porphyrosiphon sp., Nostoc sp.). The photosynthetic activity of the three mats is comparable to that observed in similar Antarctic communities, yet relies on the local factors where they grow. The soil and pond mats were distributed over moist soils and at the bottom of ponds. Unlike the stream mat, these two mats exhibited the sheath pigment scytonemin, a higher content of exopolymeric substances (EPS) and some elemental disequilibrium (C:N:P), thus suggesting great environmental stress. The areal carbon uptake (mainly through oxygenic photosynthesis) in the three mats ranged from 2.7 to 4.2 μg C cm-2 h-1, being these higher in the stream mat. The profiles with microelectrodes showed maximum photosynthetic activity at the sub-surface layers, which moreover revealed a more balanced stoichiometry than at the surface layers. Nitrogen uptakes also varied among mats. N2 fixation only occurred in the mats from the plateau, and was notably higher in the soil mats. In contrast, the areal assimilation rates of combined forms (nitrate and ammonium) were higher in the stream mat. The other benthic communities we studied are the different phototrophic biofilms which flourish in streams. Our study particularly focused on several biofilms revealing a restricted distribution within a waterfall that formed a canyon downstream. We observed up to five different communities there, whose position responded to the selective stresses exerted by stream flow and moisture. In our opinion, there is a trade-off between water current (i.e., water renewal) and nutrient availability to explain the biofilms distribution. Accordingly, the exopolymeric substances (EPS) content, stoichiometry (C:N:P) and pigment composition of the biofilms demonstrated a distinct nutritional status. Occurrence of functionally competent biofilms of chlorophytes (Ulothrix sp.) was restricted to the central stream channel, which likely indicates adaptation to faster flow events. The communities dominated by cyanobacteria (Oscillatoria spp., Phormidium cf. autumnale, Leptolyngbya sp.) were, in contrast, more diverse and appeared in a wide range of environmental conditions. The dominant diatoms in these biofilms were Fragilaria capucina s.l., Nitzschia cf. gracilis, Nitzschia inconspicua, Chamaepinnularia gerlachei, Planothidium delicatulum y Gomphonema sp., being the later related with biofilms subjected to higher drought stress. The areal photosynthetic rates (mainly oxygenic) in the biofilms ranged from 0.7 to 3.4 μg C cm-2 h-1. Greater activities were comparable to those observed in other maritime Antarctic locations, whereas the lower values obtained in the more stressed biofilms fall in the same range observed for the cyanobacterial communities of the continental region. We also assessed if the competitive interactions based on resource utilisation could explain the structure of the photosynthetic community in the microbial mats. Thus, a fully factorial nitrate and phosphorus addition experiment was conducted with the stream mat, which revealed that phosphorus fertilisation favours the growth of non-heterocystous cyanobacteria in relation to diatoms. Our experimental handling did not generate a detectable accretion of mat; however, the balanced nutrients availability improved phototrophic activity. As far as we know, this is the first attempt to study the effects of inorganic nutrient additions on the structure and function of an Antarctic microbial mat. Our findings demonstrate that biotic interactions may play a key role in structuring these aquatic food webs despite strong physical control, at least when physical stressors are temporarily relaxed. This idea contrasts partly with the general contention that ecosystems in extreme environments lack biotic control. We have certainly observed that ice dynamics and water column stability greatly explain plankton succession; nevertheless, community size structure and the occurrence of metazoan zooplankton may also explain plankton dynamics. As regards microbial mats, our enrichment experiment with stream mats also suggests that a shift in regional nutrients dynamics might alter the metabolic equilibrium of these microbial communities. The freshwater ecosystems at this site are not affected by direct anthropic stressors, but are disturbed by climate or natural eutrophication processes. Accordingly, they can be effective sentinels for climate change as they are highly sensitive and integrate information about the processes occurring in the catchment. Monitoring and experimental activities should then continue in Byers to conduct an in-depth study into the consequences of climate change on both local and global scales. The knowledge acquired by this limnological study furthermore stresses the idea of establishing this Antarctic location as an international reference site for research into polar regions.