Correspondence: Emilio O. Casamayor, Group of Limnology-Department of Continental Ecology, Centre d'Estudis Avançats de Blanes, CEAB-CSIC, Accés Cala St. Francesc, 14, E-17300 Blanes, Girona, Spain. Tel.: +34 972 336 101; fax: +34 972 337 806; e-mail: email@example.com
The bacterioneuston (bacteria inhabiting the air–water interface) is poorly characterized and possibly forms a unique community in the aquatic environment. In high mountain lakes, the surface film is subjected to extreme conditions of life, suggesting the development of a specific and adapted bacterioneuston community. We have studied the surface film of a remote high mountain lake in the Pyrenees by cloning the PCR-amplified 16S rRNA gene and comparing with bacteria present in underlying waters (UW), and airborne bacteria from the dust deposited on the top of the snow pack. We did not detect unusual taxa in the neuston but rather very common and widespread bacterial groups. Betaproteobacteria and Actinobacteria accounted for >75% of the community composition. Other minor groups were Gammaproteobacteria (between 8% and 12%), Alphaproteobacteria (between 1% and 5%), and Firmicutes (1%). However, we observed segregated populations in neuston and UW for the different clades within each of the main phylogenetic groups. The soil bacterium Acinetobacter sp. was only detected in the snow–dust sample. Overall, higher similarities were found between bacterioneuston and airborne bacteria than between the former and bacterioplankton. The surface film in high mountain lakes appears as a direct interceptor of airborne bacteria useful for monitoring long-range bacterial dispersion.
The air–water interface possibly represents a unique ecosystem in the aquatic environment. The physicochemical properties of the associated hydrophobic surface film are different from those found in the atmosphere and the water column. This interface has been described as an exposed habitat subject to environmental stresses such as solar radiation in the UV and visible spectra, toxic substances such as heavy metals and organic pollutants, and a wide range of temperature fluctuations (e.g. Norkrans, 1980; Maki, 1993; Agoguéet al., 2004 and references therein). This fact led to the belief that a specific surface bacterial assemblage – the bacterioneuston – could be present at such an interface, holding a few adapted species. Therefore, the air–water interface has often been considered as an extreme environment for microorganisms that may contain unusual species and new taxa (Maki, 2002).
Microbial neuston may also contain organisms that migrate from the benthal or pelagial areas, or that enter the interface via the atmosphere or inflowing watersheds. They may also show specifically adapted alterations in specific gravity, positive phototaxis, and enhancement of growth rates in surface films as compared with the underlying bulk water (for a review, see Maki & Hermansson, 1994). Several studies have been dealing with specific activities and abundances in the neuston for decades (e.g. Dietz et al., 1976; Hermansson & Dahlback, 1983; Maki & Remsen, 1989; Joux et al., 2005; Kostrzewska-Szlakowska, 2005). However, the bacterial diversity in the neuston is poorly known. It has been studied to some extent in the sea with genetic methods (Agoguéet al., 2005a; Franklin et al., 2005; Cunliffe et al., 2008) and in some eutrophic and mesotrophic low-land lakes using mostly traditional microbiological methods (Pladdies et al., 2004; Kalwasinska & Donderski, 2005; Kostrzewska-Szlakowska, 2005). In marine environments, the air–sea interface represents a natural source of new culturable bacterial genera (Agoguéet al., 2005a), some of which have shown high resistance to UV (Agoguéet al., 2005b). However, the lack of consistent differences between the 16S rRNA genetic fingerprint from the surface microlayer and underlying waters (UW) suggests that the presence of a stable neustonic bacterial community is not a common trait in the often windy coastal marine environment (Agoguéet al., 2004, 2005a). In freshwater lakes, in turn, specific bacterial communities are formed at the air–water interface that are most evident during prolonged periods of calm weather. A typical member of this community is Nevskia ramosa (Famintzin, 1892), which forms conspicuous two-dimensional colonies. It has been shown to strongly enrich in the neuston but also to occur as single cells in the UW under other conditions (Babenzien, 1989; Pladdies et al., 2004). Because of the viscosity and surface tension, the air–water interface could have characteristics similar to a solid surface for a bacterium (Sieburth, 1983).
In the present work, we have explored whether or not bacteria entering the interface via the atmosphere, i.e. airborne bacteria, can be an important source of neustonic bacteria. For this study, we have selected a remote high mountain lake located at 2240 m altitude in the Central Pyrenees (Lake Redon, Spain) in a rocky landscape and with very little influence of its catchment area. High mountain lakes are remote pristine systems difficult to reach and with no local anthropogenic influences. Their location determines a number of environmental conditions that are considered extreme for life (Catalan et al., 2006). We studied the bacterioneuston community present in the surface film by cloning the 16S rRNA gene as compared with clone libraries from UW (0.5 m depth), and from airborne bacteria (dust) deposited on the top of the snow pack covering the lake and the catchment area. Generation of atmospheric aerosols and remote dust deposition has increased considerably in the last few years (Prospero & Lamb, 2003; Moulin & Chiapello, 2006; Neff et al., 2008), and the neuston from high mountain lakes can be a very convenient model system to test some of the perturbations linked to Global Change and bacterial dispersion.
Materials and methods
Lake description and sampling
Lake Redon (named Lake Redó in former publications: see Medina-Sánchez et al., 2005; Catalan et al., 2006) is an oligotrophic high mountain lake located in the Central Pyrenees, Spain (42°38′N, 0°46′E, 2240 m above the mean sea level). The lake has a surface area of 24 ha, a maximum depth of 73 m, and a mean depth of 32 m. Lake samples were obtained from a fixed sampling platform located on the deepest part of the lake, and benthic influence was thus null on collected samples. The total catchment area is 153 ha, mostly of a rocky landscape with small meadows, and the water residence time is 4.2 years. The lake and the catchment area are usually covered by snow for 6–7 months of the year, normally from December to the end of June. No inflowing rivers feed the lake, which remains quite isolated because of its headwater nature. More details and a lengthy description of the biology and limnology of the lake carried out for the last 25 years can be found in a recent review (Catalan et al., 2006).
The bacterioneuston was collected from the upper c. 400 μm of the surface film with a nylon screen sampler (Agoguéet al., 2004; Auguet & Casamayor, 2008) from three different sites across the lake in August 2004, and pooled together. A sample from UW (0.5 m depth) was collected at each sampling site and pooled together as well. Previously, airborne bacteria deposited on the top of the snow pack covering both the lake and the catchment area were collected after a Saharan dust deposition event in June 2004 (http://www.calima.ws). Surface snow and dust was collected from the first centimeter of three randomly selected sites placed on the frozen lake (c. 1 m2 each), mixed and melted for further processing. Table 1 shows the total volumes processed, bacterial abundances by DAPI counts, and some additional data of the three different sampling sites. Additionally, the doses of UV radiation in Lake Redon are high specially in the most damaging part (UV-B: 280–320 nm), and the surface film experiences higher doses than UW (for more details, see fig. 5 in Catalan et al., 2006).
Table 1. Background data for samples in Lake Redon and predicted values of coverage, SChao1 and SACE in the 16S rRNA gene clone libraries
Volume sampled (mL)
DAPI (cells mL−1) × 105
Chl a (μg L−1)
DOC (mg L−1)
NA, not available. UW (0.5 m depth). DOC, dissolved organic carbon.
DNA extraction, PCR amplification, and 16S rRNA gene clone libraries
Water samples were prefiltered through a 20-μm net and then filtered on 0.2-μm pore polycarbonate membranes. Filters were incubated with lysozyme, Proteinase K, and sodium dodecyl sulfate in lysis buffer (40 mM EDTA, 50 mM Tris, pH 8.3, and 0.75 M sucrose), and phenol extracted as described previously (Dumestre et al., 2002). The bacterial 16S rRNA gene was PCR amplified with primers 27 forward (5′-AGA GTT TGA TCM TGG CTC AG-3′) and 1492 reverse (5′-GGT TAC CTT GTT ACG ACT T-3′) at 48 °C annealing temperature as previously described (Ferrera et al., 2004). PCR products were purified with the QIAquick® PCR Purification kit (Qiagen) and cloned with the TOPO TA® cloning kit (Invitrogen) following the manufacturer's instructions. Up to 500 clones equally distributed among the different samples were randomly selected and digested with the enzyme HaeIII, and restriction fragment length polymorphism patterns (operational taxonomic units, OTUs) were observed in 3% low-melting-point agarose gels. Two clones were selected from each OTU for sequencing. Clone library coverage (C) was calculated according to the following equation: C=1−(n/N) × 100, where n is the number of unique clones and N is the total number of clones examined (Ravenschlag et al., 1999), and richness estimators SACE and SChao1 were calculated as reported by Kemp & Aller (2004).
Sequencing and phylogenetic analysis
Sequencing reactions were carried out using external facilities (http://www.macrogen.com), and 16S rRNA gene sequences were submitted for preliminary identification to blastn search (http://www.ncbi.nih.gov/BLAST). Chimerical sequences were identified by check_chimera (Maidak et al., 2000) and by visual inspection of sequence alignments. Sequences were aligned with the arb program package (http://www.arb-home.de) and were inserted into the optimized and validated tree available in arb by the maximum-parsimony criterion and a special arb parsimony tool that did not affect the initial tree topology. Sequence data were submitted to the EMBL database, and accession numbers are indicated in Figs 1–4.
We obtained >120 different 16S rRNA gene sequences from the different libraries, with a coverage value >85% in all three cases (Table 1). Coverage was lower in the neuston and snow–dust (c. 85%) than in UW (96%). The richness estimators SACE and SChao1 consistently showed higher richness in the snow–dust and neuston libraries than in UW (Table 1), and the ratio observed/predicted phylotypes for each estimator was twice in UW than in the other two libraries. Betaproteobacteria were the predominant group in the three libraries reaching 67%, 71%, and 82% of the total recovered sequences in the dust–snow, neuston, and UW, respectively (Table 2). Actinobacteria were also abundant and, overall, sequences related to Betaproteobacteria and Actinobacteria accounted for >75% of the community composition detected in the three samples. Other minor groups were Gammaproteobacteria (between 8% and 12%), Alphaproteobacteria (between 1% and 5%), and Firmicutes (1% in the neuston, Table 2).
Table 2. Relative abundances (%) for the recovered clones in the different clone libraries
HGC, high GC content; LGC, low GC content. Bold indicates total % per group.
Beta-I cluster-GSK16 group
Beta-I cluster-Rhodoferax group
Total HGC Actinobacteria
Total LGC Firmicutes
Phylogenetic analysis of the 16S rRNA gene showed that most of the Betaproteobacteria fitted into three previously described freshwater clusters, Beta-I (most of them), Beta-II, and Beta-III (Glöckner et al., 2000; Zwart et al., 2002). We observed a segregation between the Beta-I subclusters GSK16 and Rhodoferax-like bacteria in agreement with the origin of the sequences (Fig. 1). Sequences from the Rhodoferax subcluster were mostly found in the snow–dust and neuston, and were close to the detection limits in UW (Table 2). Conversely, sequences affiliated to the GSK16 subcluster were, in relative terms, more abundant in UW (up to 69% of the total clones analyzed) than in the other sites. These differences in spatial distribution suggest different ecologies and/or physiologies for these two closely related Betaproteobacteria. We did not detect sequences from the Beta-II and Beta-III clusters in the snow–dust sample, but we observed Beta-II bacteria (Polynucleobacter-like cluster) to be enriched in the neuston as compared with UW, and the opposite trend for the Beta-III bacterial sequences (Table 2, Fig. 2), suggesting again different ecologies and/or physiologies for each group. Interestingly, we observed sequences from airborne bacteria reported in recent studies carried out in Boulder, CO (Fierer et al., 2008), and Dijon, France (P.A. Maron, unpublished data; see accession number AY632053 in GenBank), to be clustered within the Rhodoferax subcluster (Fig. 1) but not in the case of Beta-II and Beta-III clusters (Fig. 2). In addition, other 16S rRNA gene clones only detected in the neuston and distantly related to Aquabacterium sp. (92% similarity in the 16S rRNA gene sequence) clustered together with airborne 16S rRNA gene clones from the same study (Fig. 2).
Actinobacteria was the second most abundant group in the dust–snow and neuston samples (c. 20%), but not in UW (up to 5%) (Table 2). All these sequences specifically clustered within the uncultured acI-A cluster (Warnecke et al., 2004) and were closely related to 16S rRNA gene sequences from other freshwater environments (Fig. 3). In turn, in the dust–snow sample, Gammaproteobacteria were mostly related to cultured strains of Acinetobacter spp. (Table 2). Acinetobacter-like sequences have also been observed in air samples from Boulder, CO (Fierer et al., 2008) and in a Saharan dust enrichment carried out in a high mountain reservoir in Southern Spain (Reche et al., 2008) (Fig. 3). Other sequences from the neuston clustered in the Moraxellaceae group, together with airborne bacterial clones from air samples (Fierer et al., 2008).
Finally, we detected sequences related to Alphaproteobacteria in all three libraries (Table 2, Fig. 4), but again showing spatial segregation. Whereas in UW the sequences were closely related toRhodopseudomonas-like cluster, in the dust–snow and mainly in the neuston, the sequences clustered in the Sphingomonas–Sphingobium group near the previously described freshwater clusters Alpha-III and Alpha-IV (Glöckner et al., 2000). Such sequences were closely related (98–99% similarity) to airborne bacterial clones (Fierer et al., 2008) and bacteria obtained after an enrichment experiment using aerial dust (see AM950231 and AM950232 in GenBank).
Comparison of the different bacterial assemblages as depicted by culture-independent 16S rRNA gene PCR amplification and cloning showed higher similarity between bacterioneuston and airborne bacteria than the former with bacterioplankton. Thus, the results presented here indicated that airborne inputs could be an important source of bacteria in the surface film of this high mountain lake affected by aerosol depositions and located above the tree-line. We did not detect, however, unusual taxa in the neuston of Lake Redon but rather very common and widespread bacterial groups. Betaproteobacteria and Actinobacteria were the most abundant groups as reported in former works for other alpine areas (Alfreider et al., 1996; Glöckner et al., 2000; Zwart et al., 2002), and cosmopolitan Gammaproteobacteria closely related to Acinetobacter spp. and a few Firmicutes were also detected. Of course, other minor populations specifically adapted to live in the neuston (e.g. Maki & Remsen, 1989) might have escaped the PCR-cloning approach with the universal primers used here. The use of either more-specific primers (e.g. Cunliffe et al., 2008) or microscopy combined with FISH may have detected specific neustonic bacteria, although such specific populations certainly are minor components within the whole bacterial assemblage present in the surface film. In addition, it should also be considered that these results may be somewhat influenced by the sampling procedure used here. The surface film is operationally defined based on the depth of the sample layer collected, which is, in turn, dependent on the sampler used (see a recent comparison in Agoguéet al., 2004, and references therein). Sampling with polycarbonate and teflon membranes offers the advantage of collecting a much thinner layer (10–50 μm) than the nylon screen used here (c. 400 μm), but the selective surface adsorption of bacteria to membranes may strongly bias the final picture (Agoguéet al., 2004, and references therein). Bacterial abundance estimations carried out in several high mountain lakes of the Central Pyrenees showed in most cases weak enrichment (i.e. ratio between abundance in the neuston vs. UW) or no enrichment at all (Auguet & Casamayor, 2008). Thus, the surface film holds a bacterial concentration in the same range as that of UW, and most of the bacteria present in the surface film seem to have their origin in airborne bacteria. However, certain groups such as the Crenarchaeota showed a tendency to be selectively enriched in the surface film of high mountain lakes, and had their origin neither in terrestrial inputs from the surrounding landscape nor in soils (Auguet & Casamayor, 2008), suggesting that they may find better conditions for growth there. The same spatial segregation in species composition was observed here for certain groups of bacteria.
In the case of Betaproteobacteria, we observed different vertical distributions within the Beta-I cluster for Rhodoferax-like bacteria and GKS16, as well as for the Polynucleobacter-like cluster (Beta-II) and the Beta-III, respectively. From previous reports in the literature, different Betaproteobacteria subgroups were differently associated with environmental variables (Lindstrom et al., 2005), and some other differences were apparent among these different Betaproteobacteria. The GSK16 subcluster contained 16S rRNA gene sequences mainly recovered from freshwater ultraoligotrophic cold environments, such as Crater Lake, subglacial environments, and alpine and nival lakes, whereas the Rhodoferax subcluster is a cosmopolitan freshwater group very abundant in humic and eutrophic lakes too (10–50% of the total bacteria, Zwart et al., 2002; Simek et al., 2005), and also detected in atmospheric samples (Fierer et al., 2008). Some members of the Rhodoferax cluster are very fast growing populations that quickly react to changes in environmental conditions but are vulnerable to protistan bacterivory, thus representing bacteria with an opportunistic strategy (Simek et al., 2005). The Beta-II also forms a cosmopolitan freshwater lineage that contains phylotypes rapidly responding to changes in available substrate levels (Burkert et al., 2003). The neuston in these ultraoligotrophic aquatic environments is the first interceptor of atmospheric depositions and it is well reported that aerosols contain mineral nutrients (nitrogen, phosphorus, and iron) and organic carbon (Morales-Baquero et al., 2006) that could support the growth of freshwater bacteria (Reche et al., 2009). Unfortunately, data on the ecology and activity for the remaining freshwater Betaproteobacteria lineages are still very scarce and difficult to link to the conditions prevailing in the surface film of high mountain lakes. Altogether, data indicate that certain groups of freshwater Betaproteobacteria could be easily dispersed by air and tend to accumulate in the very top surface of high mountain lakes, whereas some others are more limited to the planktonic realm. Certainly, different ecologies and/or physiologies are involved in the different freshwater Betaproteobacteria groups detected, also affecting their vertical distribution, that deserve further research at a higher phylogenetic resolution detail.
Actinobacteria are particularly abundant in alpine lakes at a high altitude (Warnecke et al., 2005), and the ac-I freshwater lineage is one of the most ubiquitous and abundant groups of lake bacterioplankton (Newton et al., 2007). Sequences from the freshwater actinobacterial lineages ac-II to ac-IV were not detected either in Lake Redon or in the dust/snow sample. Intriguingly, none of the bacterial 16S rRNA gene sequences obtained from air samples in Boulder, CO, matched within this ac-I cluster (Fierer et al., 2008). Actinobacteria, formerly called Actinomycetes, form a major bacterial population in soil and, theoretically, such bacteria might be entering the lakes from the catchment soils as dormant cells. It has been shown that Actinobacteria from different lineages of the ac-I clade are active and very successful members of the planktonic assemblages in mountain lakes (Warnecke et al., 2005). In addition, Actinobacteria could also be successfully transported for long distances in the troposphere during dust outbreaks. Massive airborne plumes of Saharan dust entering the Mediterranean region are very common and show clear seasonal and climatic patterns (Moulin et al., 1997; Rodriguez et al., 2001), suggesting continuous feeding with allocthonous bacteria. Most of the particles in the aeolian dust are <6 μm in diameter because anything larger tends to fall out fairly quickly. Particles <6 μm, however, can remain in the atmosphere for days, weeks, or even months (Pöschl, 2005). These airborne particles have a shadow effect against high UV radiation in the troposphere and may even protect bacteria against complete desiccation conditions (Pöschl, 2005). Many Actinobacteria strains are pigmented and spore forming. It has been suggested that pigments are important for the resistance of bacteria to solar radiation (Hermansson et al., 1987), although evidences have also shown that pigmented marine bacteria are not more resistant than the nonpigmented ones to UV (Agoguéet al., 2005b). Spores, in turn, are well known as a highly resistant mechanism to unfavorable environmental conditions and can remain viable under harsh conditions for extended periods of time, promoting long-term bacterial survival and dispersion. Finally, the high GC content has also been reported as a protection mechanism against the harmful effects of UV (Singer & Ames, 1970). Although more studies are needed to draw further conclusions, it is possible that airborne freshwater Actinobacteria could be closely related to dust outbreaks. Curiously, Firmicutes (low GC content Gram-positive) are also able to produce endospores and are very resistant to desiccation. They were rarely detected in the dust/snow sample (1% of the total clones) but not in Lake Redon. In fact, Firmicutes are usually not found in freshwater bacterioplankton. Alternatively, freshwater Actinobacteria may grow in the neuston and on the snow surface very efficiently using the nutrients carried by aerosols. Altogether, the occurrence of Actinobacteria in the neuston does not necessarily imply that these bacteria are actively growing and should be considered true bacterioneuston (Maki, 1993), but they certainly are equipped with the best strategies for airborne colonization.
Finally, Alpha- and Gammaproteobacteria are usually poorly represented in freshwater bacterioplankton and our results agreed with this general pattern. Studies carried out in the ice and snow of the Himalayan area (Liu et al., 2006a, b; Zhang et al., 2006) obtained similar populations of Sphingomonas-like and Acinetobacter-like clusters as we have obtained in the Pyrenean area. Most species within the Sphingomonas–Sphingobium cluster are adapted to deal with intense solar radiation and desiccation, also surviving under low-nutrient or starvation conditions. Acinetobacter is a gammaproteobacterium ubiquitous in soil and water with high ability to remain viable under dry conditions and easily transported by air (Wendt et al., 1997). Later, however, Acinetobacter spp. detected in the dust samples were not able to successfully colonize either the neuston or the plankton in the Lake Redon. We have recurrently found Acinetobacter-like sequences in similar experiments carried out with dust deposition in Sierra Nevada (Southern Spain, Reche et al., 2009) and in the Pyrenees (Hervas & Casamayor, unpublished data), and Acinetobacter sp. has been reported on the snow in the Mt Everest region as well (Liu et al., 2006a), suggesting that this bacterium might benefit from the protection against high UV radiation conferred by airborne microscopic particles among other survival abilities.
Overall, the surface film of high mountain lakes appears as a direct interceptor of airborne bacteria where microorganisms probably find quite different environmental and ecological conditions than in UW. Catchments in high mountain lakes are relatively small compared with low-land lakes, and, as a consequence, atmospheric loadings tend to determine the water characteristics significantly (Psenner, 1999; Catalan et al., 2006; Pulido-Villena et al., 2006), and, probably, the bacterial community composition too, although not all cells reaching this layer will be able to survive and grow under in situ conditions. During the melting process of the snow (mixed with dry deposition of particles) accumulated on top of the lake and along early summer, the lake experiences a continuous washout (Catalan et al., 2006), including those bacteria carried on dust particles. On the other hand, dust outbreaks reaching South Europe typically occur in spring and summer. Thus, a continuous feeding with new waves of airborne bacteria is expected along this period. Generation of atmospheric aerosols and remote deposition is an increasing process linked to some components of the Global Change. The neuston from high mountain lakes can act as sentinels to follow those species entering remote areas with the potential to colonize new environments and induce changes in the ecosystem functioning (Kellog & Griffin, 2006). The fate of these bacteria needs to be, therefore, carefully evaluated.
This research was initially funded by the grant ECOSENSOR BIOCON04/009 from the Fundación BBVA and further by the project AERBAC 079-2007 from the Ministerio de Medio Ambiente-Red de Parques Nacionales, and was carried out in the Limnological Observatory of the Pyrenees. We are grateful to the Authorities of AiguesTortes and Estany de St. Maurici National Park for sampling facilities and continuous support. A.H. was granted a PhD scholarship from the BBVA Foundation to carry out this work, and E.O.C. by the Programa Ramón y Cajal from the Spanish Ministerio de Educación y Ciencia and FEDER.