Temporal Changes and Altitudinal Distribution of Aerobic Anoxygenic Phototrophs in Mountain Lakes

Abstract

Aerobic anoxygenic phototrophs (AAPs) are bacteriochlorophyll a-containing microorganisms that use organic substrates for growth but can supplement their energy requirements with light. They have been reported from various marine and limnic environments; however, their ecology remains largely unknown. Here infrared epifluorescence microscopy was used to monitor temporal changes in AAPs in the alpine lake Gossenköllesee, located in the Tyrolean Alps, Austria. AAP abundance was low (10³ cells ml⁻¹) until mid-July and reached a maximum of ∼1.3 × 10⁵ cells ml⁻¹ (29% of all prokaryotes) in mid-September. We compared the studied lake with other mountain lakes located across an altitudinal gradient (913 to 2,799 m above sea level). The concentration of dissolved organic carbon and water transparency seem to be the main factors influencing AAP abundance during the seasonal cycle as well as across the altitudinal gradient. While the AAP populations inhabiting the alpine lakes were composed of intensely pigmented large rods (5 to 12 μm), the lakes below the tree line were inhabited by a variety of smaller morphotypes. Analysis of pufM diversity revealed that AAPs in Gossenköllesee were almost exclusively Sphingomonadales species, which indicates that AAP communities inhabiting alpine lakes are relatively homogeneous compared to those in low-altitude lakes.

INTRODUCTION

The discovery of photoheterotrophic bacteria in the upper ocean has attracted the interest of aquatic microbiologists over the past decade (1). One group of these organisms, the aerobic anoxygenic phototrophs (AAPs), harvest light energy using bacteriochlorophyll a (BChl a)-containing reaction centers to supplement their predominantly chemoheterotrophic metabolism (2, 3). AAPs represent 1 to 11% of the total prokaryotes in the euphotic zone of the world's oceans (4, 5, 6). In more eutrophic environments such as shelf seas and estuaries, they can comprise up to 34% of the bacterial community (7, 8). Marine AAPs represent a phylogenetically heterogeneous group which includes different clades of Alpha-, Beta- and Gammaproteobacteria (9, 10).

In contrast to marine AAPs, their freshwater counterparts have received until now little attention. The first AAP freshwater species was isolated by Yurkov and Gorlenko (11, 12) from alkaline cyanobacterial mat surfaces. Later, several other AAP strains were isolated from various limnic habitats (13–16). Culture-independent approaches were first applied by Waidner and Kirchmann (17), who found two bacterial photosynthesis gene clusters in a fosmid library constructed from the Delaware River, USA. The presence of AAPs in Swedish lakes was documented using bchL clone libraries (18). Later, an analysis of pufM clone libraries showed that the AAP community inhabiting lakes in Germany is formed of a broad variety of alpha- and betaproteobacterial species (19). Photosynthesis genes signalizing AAP species were also found in several single-cell genomes collected from freshwater lakes in Maine and Wisconsin (20).

The first systematic surveys of AAP abundances in various limnic systems in central Europe revealed that these organisms represent a significant part of the microbial community in many oligotrophic and mesotrophic lakes (21, 22). Interestingly, the greatest abundances were found in the acidified mountain lakes Plešné and Čertovo in the Šumava Mountains (Böhmerwald; 1,090 and 1,028 m above sea level [a.s.l.]), where during the summer maximum AAPs formed more than half of the bacterial biomass (21).

In general, mountain lakes in temperate regions can be divided into two main groups. Alpine lakes are located above the tree line, and their catchments are formed of bare rocks and poorly developed soils. They are characterized by harsh conditions, such as high levels of UV radiation, low temperatures, and low concentrations of dissolved organic carbon (DOC) and nutrients (23, 24). The second group, subalpine lakes, are located below the tree line, and their catchment areas are forested. Compared to alpine lakes, they have higher DOC and nutrient concentrations and lower water transparency as well as higher water temperatures (24, 25).

The ecological drivers of AAP distribution in aquatic systems are not yet entirely understood. It was hypothesized that the ability of these bacteria to use light energy should be beneficial for their photoheterotrophic growth in high-light but low-nutrient environments (26). Clear alpine lakes are a good example of such a habitat, where oligotrophic conditions and high-light fluxes may favor the growth of photoheterotrophic organisms (15). As there exists no systematic study on AAPs from alpine lakes, we monitored their dynamics in the alpine lake Gossenköllesee during the ice-free season. In addition, we assessed the abundances of AAPs in several mountain lakes located at different altitudes. We hypothesized that AAP abundance in this type of lake should mostly reflect changes in nutrients and solar radiation, though other factors such as water temperature and chlorophyll concentration should also modulate their distribution.

(This work was conducted by E. Holzer in partial fulfillment of the requirements for a M.Sc. degree from the University of Innsbruck, Austria, 2011.)

MATERIALS AND METHODS

Study sites and sampling.

The temporal study was conducted in the clear alpine lake Gossenköllesee (GKS), located at 2,417 m above sea level in the Tyrolean Alps, Austria (47°13′N, 11°00′E), from June to October 2009. Water samples were collected in the central part of the lake at five depths (0, 2, 4, 6, and 8.5 m) with a frequency of two or three samplings per month. The lake has a surface area of 1.7 ha and a maximum depth of ∼9.9 m. It has a dimictic and holomictic thermal regime. Its catchment area is composed of crystalline bedrock and is covered with a poor soil layer, which results in a very limited input of dissolved organic matter (DOM). Thus, DOC concentrations in GKS are at the lower end of the scale (10 to 54 μM) known for lakes (27). As a consequence of its oligotrophic condition and cold water temperatures, the water column is in most cases oxygen saturated (28).

For comparison, we surveyed eight more mountain lakes (Table 1) located in the Tyrolean Alps between 10 and 14 August 2009 (Table 2). Five of the selected lakes were located above the tree line (ca. 2,000 m a.s.l.) with poorly developed soils in their catchments (Schwarzsee [SOS], Rotfelssee [ROT], Oberer Plenderlesee [OPL], and Geirneggsee [GEI]) or with catchments partially covered by alpine meadows (Drachensee [DRA]). Three lakes (Seebensee [SEE], Piburgersee [PIB], and Achensee [ACH]) were located below the tree line and thus had forested catchments. All lakes have oxic conditions in the water column, except for PIB, which has anoxic conditions below the 20-m depth.

Table 1.

Altitude, latitude, longitude, maximum lake depth, lake area, catchment area, and characteristics of catchment area of the studied lakes^a

Characteristics of catchment areas and names of lakes	Lake code	Altitude (m a.s.l.)	Latitude	Longitude	Maximum depth (m)	Lake area (ha)	Catchment area (ha)
Exposed rock
Schwarzsee	SOS	2,799	46°57′	10°56′	18	3.5	18
Rotfelssee	ROT	2,485	47°14′	11°00′	5.5	0.9	33
Gossenköllesee	GKS	2,417	47°13′	11°00′	9.9	1.7	30
Geirneggsee	GEI	2,410	47°13′	11°00′	1	0.4	ND
Oberer Plenderlesee	OPL	2,344	47°12′	11°02′	7.5	2.1	97
Meadows
Drachensee	DRA	1,874	47°02′	10°56′	24	4.5	188
Forests
Seebensee	SEE	1,650	47°21′	10°56′	14	6.4	ND
Achensee	ACH	929	47°27′	11°42′	133	680	10500
Piburgersee	PIB	913	47°11	10°53	24.6	13.4	265

^aND, no data.

Table 2.

Basic physicochemical characteristics and total bacterial counts for surface waters of lakes studied in August 2009^a

Lake	Date	Temp (°C)	pH	Chl a (μg liter−1)	DOC (μM)	TDP (μM)	Kd PAR (m−1)	Total bacteria (105 cells ml−1)
SOS	13 Aug	9.4	6.25	1.18	33.3	0.022	0.215	3.3
ROT	12 Aug	11.1	7.35	1.46	26.7	0.053	ND	2.2
GKS	12 Aug	12.9	7.24	1.04	31.7	0.016	0.183	2.8
GEI	12 Aug	13.4	7.40	1.94	25.8	0.034	ND	1.6
OPL	10 Aug	10.0	7.08	0.69	25.8	0.034	ND	4.0
DRA	11 Aug	12.6	8.12	0.69	31.7	0.022	0.179	3.6
SEE	11 Aug	11.2	8.33	1.18	42.5	0.038	0.211	18.0
ACH	14 Aug	19.2	8.41	1.44	116.7	0.047	ND	10.8
PIB	13 Aug	22.2	8.15	2.43	191.7	0.075	0.437	25.0

^aND, no data.

Lake water was collected with a 5-liter Schindler-Patalas sampler from three to eight depths, depending on lake depth and water column thermal structure. The selection of sampling depths was based on profiles of chlorophyll fluorescence and light. The water samples were transported to the laboratory in plastic carboys and processed within 4 h after sample collection.

Physicochemical parameters.

Water temperature was measured immediately after sample collection using a glass thermometer (±0.1°C) placed inside the sampler and additionally with the temperature sensor of a PUV-501B profiling radiometer (Biospherical Instruments, Inc., San Diego, CA) used for downwelling photosynthetically active radiation (PAR) measurements. The diffuse attenuation coefficient for downwelling PAR (K_d) in a water column was determined from the slope of linear regression of the natural logarithm of PAR versus depth. Three parallel profiles were made on each sampling date, and the mean K_d was calculated. Measurement of pH values was done with a WTW pH meter.

Subsamples for DOC and total dissolved nitrogen (TDN) analyses were filtered immediately after sampling through a double-precombusted (4 h at 450°C) GF/F filter (Whatman) placed inside a stainless steel syringe holder. Filters were rinsed first with Milli-Q water and then with the sample. The filtrate was collected in combusted glass bottles (Schott, Germany), acidified with HCl to pH 2, and stored in the dark at 4°C until further analysis (within 48 h). DOC concentrations (as nonpurgeable organic carbon) were measured with a total organic carbon analyzer (Shimadzu TOC-Vc series) equipped with a TNM-1 module. Both parameters were detected simultaneously after combustion and catalytic oxidation of the injected sample. Three to five injections were analyzed for each sample. A consensus reference material (CRM) analysis for DOC (batch 5 FS-2005; 47.5 μM C) provided by the Rosenstiel School of Marine and Atmospheric Science, Division of Marine and Atmospheric Chemistry, University of Miami, was run in parallel. Results differed from the CRM given value by 5%, and the coefficient of variation was better than 2%. For total dissolved phosphorus (TDP) analysis, water samples were placed in polyethylene bottles (acid washed with HCl, Milli-Q water, and sample water) and transported to the laboratory at in situ temperatures, where they were filtered through a glass fiber filter (GF/F; Whatman) on the same day (within 6 h) and stored at 4°C until further analysis within 24 h. The TDP concentration was determined spectrophotometrically using the molybdate method after digestion with sulfuric acid and hydrogen peroxide (29).

Chlorophyll-a (Chl a) was determined spectrophotometrically in acetone extracts (30). Briefly, a known volume of lake water from each depth (2 to 3 liters) was filtered through a glass fiber filter (25 mm; Whatman GF/F) and the pigments from phytoplankton were extracted in 90% alkaline acetone. The absorbance was measured in a double-beam spectrophotometer (Hitachi U-2001) between 400 nm and 750 nm. After the measurement, the sample was acidified with 1 M HCl to pH 2, and after 5 min, the absorbance was measured again to account for the interference of phaeopigments.

Microscopy.

For microscopic analysis, water subsamples were fixed with 0.2 μm prefiltered formaldehyde (2% final concentration), stored at 4°C, and processed within 3 weeks. Infrared (IR) epifluorescence microscopy was done with an Olympus BX51TF fluorescence microscope equipped with an Olympus Universal Plan Apochromat 100×/1.35 oil objective lens and a black and white CCD F/View II camera (7). First, total 4′,6-diamidino-2-phenylindole (DAPI)-stained bacteria were recorded in the blue part of the spectrum (100- to 200-ms exposure time). Then the Chl a autofluorescence image (0.5- to 1-s exposure time) and the IR emission image (15- to 35-s exposure time) were captured in the same field of view. For exact counts of BChl a-containing bacteria, the contribution of Chl a-containing organisms was subtracted from the obtained IR images. Approximately 10 randomly selected microscopic fields were recorded and manually analyzed with the aid of AnalySiS software (Soft Imaging System GmbH, Germany).

The composition of the bacterial community was quantitatively assessed by the catalyzed reporter deposition fluorescence in situ hybridization (CARD-FISH) method as described in references 31 and 32. Five different group-specific 5′ horseradish peroxidase (HRP)-labeled oligonucleotide probes (Thermo-Hybrid, Germany) were used: for the Bacteria domain, EUB338 I-III; for Alphaproteobacteria, ALF986; for Betaproteobacteria, BET42a; for Cytophaga-like bacteria, CF319a; and for Actinobacteria, HGC69a. Gammaproteobacteria were not targeted because this group is not abundant in Gossenköllesee (32). Cells were counted using a Zeiss Axioplan epifluorescence microscope in 20 different randomly selected microscopic fields. For each field, a number of DAPI-positive cells and a number of hybridized cells were determined. At least 400 DAPI-stained cells were counted per filter segment, but for most samples 600 to 700 cells were counted.

Environmental pufLM gene clone library construction.

Water samples for DNA extraction were collected from 0-m, 3-m, and 6-m depths, pooled, and filtered (300 to 600 ml) onto a 0.2-μm polycarbonate filter (diameter, 47 mm; Nuclepore; Whatman Ltd., USA). Plankton genomic DNA was extracted using phenol-chloroform extraction (19). The partial pufLM genes were PCR amplified from the environmental DNA samples using degenerated primers described previously (9). The obtained PCR products were purified using a GenElute PCR clean-up kit (Sigma) and cloned in competent Escherichia coli DH5-α cells according to the manufacturer's protocol for the pGEM-T Easy vector system (Promega). Single colonies positive for ampicillin resistance, distinguished on the basis of blue-white screening, were picked and used for screening target inserts with the vector-specific primers SP6 and T7. The positive clones were sequenced (Macrogen Inc., South Korea), and the obtained sequences were manually checked. The sequences were imported into an ARB database (http://arb-home.de) and aligned. A distance matrix was calculated with Mothur (33). The resulting matrix was used to group all sequences to operational taxonomic units (OTUs), and the retrieved OTUs were used for phylogenetic analysis. Based on an ∼700-nucleotide sequence length, a phylogenetic backbone tree was calculated, applying maximum likelihood criteria using the RAxML method integrated with ARB. The tree topology was verified by using 1,000 bootstrap replications. Sequences shorter than 700 nucleotides were added later, applying maximum-parsimony criteria using the add-by-parsimony tool of ARB.

Statistical analysis.

The influence of environmental parameters on AAP abundance was tested by redundancy analysis (RDA) with forward selection. The following environmental parameters were included as explanatory variables for the seasonal study: water temperature, Chl a, DOC, TDN, and TDP concentrations, and K_d PAR. The species data were centered and standardized prior to analysis. In the case of the altitudinal-gradient survey, we used altitude, water temperature, Chl a, DOC, and TDP concentrations, K_d PAR, and pH as explanatory variables. To determine the contributions of specific environmental variables and their significance, we used forward selection with the Monte Carlo permutation test (499 unrestricted permutations). All computation was done with Canoco for Windows 4.5 (34).

Nucleotide sequence accession numbers.

Representative sequences for each OTU were submitted to NCBI GenBank under accession numbers JN715070 through JN715076.

RESULTS

Seasonal survey.

Intensive seasonal sampling was conducted on GKS during the entire ice-free period in 2009. After ice-out, the temperature increased in the upper water layers until it reached its maximum (14.8°C) on 24 August, whereas at a depth of 8.5 m, the maximum was measured at the end of September (Fig. 1A). The upper mixed layer increased from 3.5 m on 8 July to a 7.8-m depth on 15 September (see Fig. S1 in the supplemental material). The rapid change in the density structure of the water column to mid-September resulted from several days of bad weather, as evidenced by low values for global radiation (see Fig. S1) and a sudden decrease in water temperature. In June the Chl a concentration stayed low (1.14 ± 0.69 μg liter⁻¹). In July a strong, deep Chl a maximum (6.1 μg liter⁻¹) developed at the bottom of the lake (Fig. 1B) but gradually decreased during the rest of the season. Total prokaryotic abundance (DAPI-stained cells) ranged from ∼2 × 10⁵ to 6 × 10⁵ cells ml⁻¹ (Fig. 1C), with the maximum observed on 23 September at an 8.5-m depth. Water transparency to PAR decreased (i.e., higher K_d values) after 8 July until the end of the study period (Table 2).

Fig 1.

Seasonal changes in temperature (A), chlorophyll concentration (B), total bacteria (C), and AAP abundance (D) in the water column of GKS.

The composition of the bacterial community was analyzed by CARD-FISH (Table 3; see also Fig. S2 and S3 in the supplemental material). At the beginning of the study, Alphaproteobacteria had the lowest relative abundance in the water column (on average, 4.2% of total bacteria), but the abundance increased up to 19% on 15 September. Actinobacteria were the most abundant group in June, but their contribution declined later in the season. The average relative abundance of Cytophaga-like bacteria in the water column was highest in June but then decreased, reaching an average of 7.6% of the total bacteria on 15 September. The maximum average contribution of Betaproteobacteria (25.8%) was found on 29 July, whereas the minimum (10.6%) was observed on 31 August.

Table 3.

Average relative abundances (% total prokaryotes) of major bacterial groups determined by FISH in GKS in 2009^a

Date (mo/day)	EUB I-III (%)	Alpha (%)	Beta (%)	CF (%)	HGC (%)
Jun. 4	85.2 ± 2.2	4.2 ± 1.6	17.5 ± 1.5	35.6 ± 1.7	24.3 ± 1.3
Jun. 18	83.9 ± 2.6	5.7 ± 1.3	21.8 ± 1.6	23.8 ± 1.3	29.4 ± 3.0
Jul. 8	69.8 ± 3.8	11.8 ± 2.2	23.8 ± 3.2	13.8 ± 1.1	18.8 ± 4.8
Jul. 29	67.2 ± 5.7	12.3 ± 2.1	25.8 ± 4.1	15.3 ± 4.2	10.4 ± 7.8
Aug. 12	59.9 ± 4.7	14.1 ± 4.4	21.2 ± 2.9	18.4 ± 4.4	1.3 ± 1.3
Aug. 24	65.3 ± 4.6	18.6 ± 6.5	24.3 ± 4.1	15.2 ± 6.2	4.0 ± 2.5
Aug. 31	55.2 ± 8.5	14.2 ± 2.8	10.6 ± 2.7	13.3 ± 9.5	0.2 ± 0.2
Sep. 15	47.2 ± 4.1	19.0 ± 4.3	13.2 ± 4.5	7.6 ± 4.1	0.5 ± 0.6
Sep. 23	50.9 ± 6.0	16.5 ± 5.7	15.2 ± 5.2	8.4 ± 5.7	1.0 ± 1.4
Sep. 30	53.2 ± 8.6	16.7 ± 4.8	15.5 ± 3.7	10.8 ± 5.2	0.4 ± 0.5

^aEUB I-III, Bacteria; Alpha, Alphaproteobacteria; Beta, Betaproteobacteria; CF, Cytophaga-like bacteria; HGC, Actinobacteria. Values represent means of five depths ± 1 SD.

AAPs were found during the entire ice-free season, but their abundances varied by two orders of magnitude (Fig. 1D). After ice-out at the beginning of June, AAP abundance was low (10³ cells ml⁻¹). During the summer, AAP abundance gradually increased, reaching its maximum on 15 September (1.3 × 10⁵ cell ml⁻¹). Except for 8 July, the AAP maximum was always observed at mid-water layers (4- to 6-m depth) (see Fig. S5 in the supplemental material). The highest relative abundance of AAPs (29% of DAPI-stained cells) was observed on 31 August at a 2-m depth, whereas the lowest was found at the beginning of the study (∼1% of DAPI-stained cells).

During the seasonal cycle, there was a positive correlation between water temperature and AAP abundance when data for the whole water column (Pearson's r = 0.607; n = 50; P < 0.0001) were considered, but there was no correlation between AAP abundance and chlorophyll concentration (Pearson's r = −0.280; P = 0.0487; n = 50). Similarly, neither the TDN nor the TDP concentration was correlated with the abundance of AAPs. Since the found Pearson's correlations were weak, we further analyzed the data set using a redundancy analysis. AAP abundance was mostly affected by the combination of K_d PAR (62.4%; P = 0.001) and DOC concentration (13.4%; P = 0.012), which explained 75.8% of its variance (Fig. 2), while all variables explained 94% of the total variance. AAP abundance was significantly correlated with the absolute abundance of Alphaproteobacteria (Pearson's r = 0.872; P < 0.001; n = 50) for the whole data set, as well as for each depth (see Fig. S4 in the supplemental material), which suggests that AAPs inhabiting GKS were mostly Alphaproteobacteria. No significant correlations were found with any other major bacterial group.

Fig 2.

Redundancy analysis using AAP abundances (AAPs) and relative contributions of the main bacterial groups as dependent variables (solid lines) and DOC, water temperature (temp), K_d PAR, and Chl a, TDP, and TDN concentrations as explanatory variables (dashed lines). The horizontal axis explains 58.1% of bacterial-group variability, with K_d PAR itself explaining 51.4% of the variability.

Altitudinal gradient.

In addition to the seasonal survey, the distribution of AAPs was surveyed in several other mountain lakes located both above and below the tree line (Table 1). The survey was conducted in mid-August 2009 to ensure that all lakes were ice-free for more than 1 month. Surface temperatures ranged from 9.4°C in SOS to 22.2°C in PIB (Table 2). In general, DOC concentrations decreased with altitude, ranging from 25.8 to 191.7 μM except for that for SOS, the highest-altitude lake, where the concentration was 33.3 μM. Epilimnetic Chl a concentrations ranged from 0.42 to 2.43 μg liter⁻¹, with the highest value found in mesotrophic lake PIB (Table 2). In most of the lakes, there was a characteristic deep Chl a maximum in the bottom layers, with concentrations ranging from 1.56 to 12.96 μg liter⁻¹.

The abundances of AAPs varied greatly across the altitudinal gradient (Fig. 3). The highest abundances were found in the lakes located at the lowest altitudes (PIB, 17.6 × 10⁴ to 34.8 × 10⁴ cells ml⁻¹; ACH, 24 × 10⁴ to 27 × 10⁴ cells ml⁻¹), whereas the numbers found in alpine lakes were significantly lower. The lowest AAP abundance was found in the highest lake, SOS (1.1 × 10⁴ to 1.9 × 10⁴ cells ml⁻¹), followed by those in OPL (1.4 × 10⁴ to 2.3 × 10⁴ cells ml⁻¹), GEI (2.3 × 10⁴ cells ml⁻¹), ROT (4.4 × 10⁴ to 4.9 × 10⁴ cells ml⁻¹), and GKS (0.8 × 10⁴ to 6.7 × 10⁴ cells ml⁻¹) (Fig. 3). The dependence of AAP abundance on lake altitude was confirmed by a strong negative correlation between these two parameters (Pearson's r = −0.934; P = 0.0002; n = 9). AAPs were also strongly positively correlated with DOC concentration (Pearson's r = 0.959; P < 0.0001; n = 9) and PAR attenuation (Pearson's r = 0.949; P = 0.003; n = 9). There was also a positive correlation of AAP abundance and temperature; however, this effect was weak (Pearson's r = 0.795; P = 0.0104; n = 9). The redundancy analysis (Fig. 4) explained 84.3% of the variabilities in total bacteria and AAP abundances. DOC concentration explained 73.3% of the data variability, and its effect was significant (P = 0.002). PAR attenuation explained 5.4% (P = 0.032) of the total model.

Fig 3.

AAP abundances within water columns (box plot) and proportional contributions to the total bacterial community in studied lakes sampled in Aug 2009. Lakes are presented according to their altitudes, from the highest (SOS) to the lowest (PIB). Mean values for DOC concentrations in water columns are marked by black dots; error bars represent ±1 SD.

Fig 4.

Redundancy analysis using total bacterial and AAP abundances (AAPs) as dependent variables (solid lines) and altitude, water temperature (temp), K_d PAR, pH, and Chl a, DOC, and TDP concentrations as explanatory variables (dashed lines). Numbers in the names of samples correspond to the sampling depths of the lakes. The whole model explained 84.3% of the variability in total prokaryotes and AAP abundances when DOC, temperature, altitude, K_d PAR, pH, and Chl a and TDP concentrations were included.

Morphological and phylogenic diversity.

Microscopic analysis identified a number of different morphotypes among AAP cells. Despite the heterogeneity, there was a clear difference between the AAP morphotypes found in alpine and subalpine lakes. Two types of rod-shaped cells with average lengths of 6 μm and 12 μm (Fig. 5) were dominant in the alpine lakes. These cells were the only morphotypes found in the four lakes located at the highest elevation (SOS, GEI, ROT, and GKS), and in OPL they still formed ca. 80% of all AAP cells. A different situation was found in lake DRA, located at the tree line. Here the large rods represented about 50% of the AAP cells, and the rest of the community was made up of oval and sickle-shaped cells (2.5 μm). In contrast to the relatively uniform morphology observed in lakes above the tree line, a variety of mostly smaller morphotypes (Fig. 5; see Fig. S6 in the supplemental material) was found in lakes located at lower altitudes. The highest diversity of morphotypes was found in PIB. Here the AAP community was dominated by large (ca. 2.5-μm) sickle-shaped (vibrioid) cells and small ovoids (1.5 μm), forming 54% and 30% of the cells, respectively. Among the less represented morphotypes found in PIB were rods, spiral cells, and ovoid cells with long stalks. Lakes ACH and SEE exhibited less morphological diversity than PIB, with the majority of morphotypes S-shaped and sickle-shaped (∼2.5 μm), which together formed around 70% of AAP cells in both lakes. Less abundant morphotypes were represented by smaller and larger ovoid cells (1.5 to 2.5 μm). We observed a statistically significant increase in the mean AAP cell size with altitude (Pearson's r = 0.93; P < 0.0001; n = 25).

Fig 5.

Scheme of main AAP morphotypes found across the altitudinal gradient.

To identify the main AAP species inhabiting GKS, we constructed a pufLM gene clone library from the GKS planktonic DNA. In total, 37 clones were sequenced, and the obtained pufM sequences were grouped into three operational taxonomic units (OTUs) based on a 98% DNA sequence similarity level. The largest OTU encompassed over 90% of the GKS sequences related to pufM sequences of Sandarakinorhabdus limnophila, Erythromonas ursincola, and Sphingomonas species (75% to 86% amino acid sequence identity), all belonging to the Sphingomonadaceae family within the alpha-4 Proteobacteria (Fig. 6). Only three sequences were related to Betaproteobacteria and formed two independent OTUs, with the Oregon Crater Lake strain HTCC528 being the closest relative (amino acid sequence identity of 76%).

Fig 6.

Maximum likelihood phylogenetic tree constructed from partial pufM nucleotide sequences. Individual OTUs detected in GKS clone libraries are marked in bold. The arrow indicates the dominant OTU that represents over 90% of the obtained clones.

DISCUSSION

AAPs were found in all the collected samples; however, their abundances varied largely during the season as well as across the altitudinal gradient. The highest abundances (3.8 × 10⁴ to 35 × 10⁴ cells ml⁻¹) were found in lakes located at the lowest altitudes (<1,000 m a.s.l.), PIB and ACH. These numbers were comparable to previously reported AAP abundances (9 × 10⁴ to 43 × 10⁴ cells ml⁻¹) in low-altitude freshwater lakes in the Czech Republic (21). The lower AAP abundances found in alpine lakes (2.7 × 10³ to 2.1 × 10⁵ cells ml⁻¹) likely reflected the overall lower trophic status of the lakes and their lower bacterial numbers. This is consistent with the fact that AAP relative abundances were similar for alpine (3 to 22%) and subalpine (4.8 to 23%) lakes (average ± standard deviation [SD], 12.8% ± 7.1%). These values are similar to estimates based on pufM gene frequencies in single-cell genomes in Wisconsin and Maine lakes (20) but significantly higher than percentages of AAPs reported for open-ocean environments (5, 6, 35, 36).

An interesting finding was the large difference observed in AAP cell morphologies in alpine and subalpine lakes (see Fig. S6 in the supplemental material). All the surveyed alpine lakes contained the same large-rod AAP morphotypes that hybridized with the alphaproteobacterial FISH probe (see Fig. S2 in the supplemental material). Analysis of pufM clone libraries from GKS suggests that the observed large rods represent the same or two closely-related organisms belonging to the Sphingomonadaceae family (alpha-4 subgroup of Proteobacteria). This family contains both heterotrophic and photoheterotrophic (AAP) species inhabiting a variety of environments, including marine and freshwater and soil (37). The affiliation with Sphingomonadaceae also confirmed that the observed cells were indeed AAP species and not any other BChl a-containing organisms. Interestingly, the cultured bacterium closest to our environmental sequences, Sandarakinorhabdus limnophila, which was isolated from a mountain lake in Germany, exhibits a similar rod-shaped morphology in its stationary growth phase (16).

The AAP community composition, as well as species richness (Chao1 index = 3.5), in GKS was different from that of previous studies in freshwater lakes. A recent study in northeastern Germany indicated that the local AAP communities were more diverse (Chao1 index = 4 to 16), with over one half of pufM sequences related to Rhodoferax fermentans (Betaproteobacteria) (19). Similarly, Rhodoferax-related sequences have been reported from other freshwater habitats such as rivers and estuaries (17, 38). One of the reasons for the frequent occurrence of Sphingomonas species in alpine lakes might be their higher resistance to UV-B radiation. Microbial communities in clear alpine lakes are exposed to high doses of UV radiation, and certain Sphingomonas species are known to be naturally more resistant to UV radiation (39) and to suffer less DNA damage (40).

A question remains regarding the main factors that influence AAP distribution. Previous studies with freshwater lakes identified that AAP numbers in various lakes correlate with total phosphorus or chlorophyll content (22), and similar results were also found in marine environments (6, 36). Interestingly, we did not observe such correlations in either the seasonal or the altitudinal survey.

In the present study, the redundancy analysis indicated that of the included parameters DOC concentration and PAR attenuation explained most of the AAP abundance variability both in the temporal study and across the altitudinal gradient. An interesting observation concerned the role of temperature. A weak correlation between AAP abundance and water temperature was found during the temporal study. A similar effect of temperature on the development of AAPs has been reported for seasonal surveys of the mountain lake Čertovo (21), the Baltic Sea (7), and a Mediterranean coastal lagoon (41).

Across the altitudinal gradient, AAP abundance was significantly correlated with lake altitude. However, this effect was likely indirect, as altitude directly influences several environmental parameters such as DOC concentration, water temperature, and intensity of UV radiation (24, 25). Statistical analysis indicated that AAP abundance was influenced mostly by DOC concentration and to a lesser extent by PAR attenuation (Fig. 4). It is also plausible to speculate that not only concentration but also changes in DOM composition and quality occurring across the altitudinal gradient (25, 27) are important for the growth efficiency of AAPs. A positive correlation between BChl a concentration and light attenuation in marine environments has previously been reported (8, 42). Whereas in the Baltic Sea data set attenuation was interpreted as mostly originating from colored DOM (CDOM) absorption (42), another study performed in Chesapeake Bay assumed that the beam attenuation mostly reflected particle-associated turbidity (8). In our case, light attenuation changes observed in GKS are mainly related to the development of phytoplankton and Chl a absorption (27), whereas across the altitudinal gradient, quantitative and qualitative changes in CDOM are the most important factor (25).

The negative relationship of AAP abundance and water transparency (light availability) is somewhat surprising, as phototrophic organisms presumably benefit from light energy. Moreover, AAP abundance was positively correlated with DOC content, which is not consistent with a previous hypothesis stating that with their capacity to utilize light AAPs should occur more frequently in nutrient-poor environments (15, 26). A possible explanation comes from the fact that AAPs are mostly organotrophic species dependent on the supply of organic substrate, whereas light has only an auxiliary function in increasing the efficiency of DOC utilization (3). Moreover, excess light may even have an inhibitory effect on the growth of AAPs (3) that can be further enhanced by its UV component. Thus, it seems that the dominant factor limiting AAP growth in alpine lakes is DOC availability, whereas light has only a minimal effect.