The Limnohabitans Genus Harbors Generalistic and Opportunistic Subtypes: Evidence from Spatiotemporal Succession in a Canyon-Shaped Reservoir

ABSTRACT

We studied the diversity of Limnohabitans using reverse line blot hybridization with Limnohabitans lineage-specific probes in the freshwater canyon-shaped Římov reservoir (Czech Republic). To examine the succession of distinct lineages, we performed (i) a study of an intensive spring sampling program at the lacustrine part of the Římov reservoir (from ice melt through a phytoplankton peak to the clear-water phase), and (ii) a seasonal study (April to November) when the occurrence of distinct Limnohabitans lineages was related to the inherent longitudinal heterogeneity of the reservoir. Significant spatiotemporal changes in the compositions of distinct Limnohabitans lineages allowed for the identification of “generalists” that were always present throughout the whole season as well as “specialists” that appeared in the reservoir only for limited periods of time or irregularly. Our results indicate that some phytoplankton groups, such as cryptophytes or cyanobacteria, and zooplankton composition were the major factors modulating the distribution and dynamics of distinct Limnohabitans lineages. The highest Limnohabitans diversity was observed during the spring algal bloom, whereas the lowest was during the summer cyanobacterial bloom. The microdiversity also markedly increased upstream in the reservoir, being highest at the inflow, and thus likely reflecting strong influences of the watershed.

IMPORTANCE The genus Limnohabitans is a typical freshwater bacterioplankton and is believed to play a significant role in inland freshwater habitats. This work is unique in detecting and tracing different closely related lineages of this bacterial genus in its natural conditions using the semiquantitative reverse line blot hybridization method and in discovering the factors influencing the microdiversity, subtype alternations, and seasonality.

INTRODUCTION

Bacteria occupy a prominent role in freshwater ecosystem processes and significantly impact lake water quality (1). However, only a small number of bacterial groups among the “typical freshwater bacteria” (2) are believed to play a significant role in inland freshwater habitats (3). The genus Limnohabitans (Betaproteobacteria), described by Hahn et al. (4) and further amended (5–7), is indisputably such a taxon.

Evidence showing the importance and crucial role of Limnohabitans bacteria, as well as other bacterioplankton lineages, is rapidly growing (e.g., see references 6 and 8 to 15). The hitherto best ecologically described fragment of the Limnohabitans genus, the RBT cluster, which was targeted by a specific fluorescence probe, R-BT065 (16), was proven to be an indispensable and abundant part of the bacterioplankton (17–19). This cluster is highly susceptible to protozoan grazing (20) and shows pronounced niche separation, even among closely related strains (21). Members of the RBT cluster were considered to be “opportunistic” strategists (15, 18), being probably largely dependent on algal exudates (17, 22, 23). They are versatile in the incorporation of low-molecular-weight (LMW) substrates with a preference for simple organic acids and monosaccharides (7, 24).

An immense step forward in the understanding of Limnohabitans phylogeny and ecology occurred with the introduction of a completely refined Limnohabitans taxonomy (7) based on a significant number of new isolates and knowledge of their strain-specific ecophysiology. This progress led to the establishment of new well-defined “lineages” that more accurately reflect the current state of the art in this field than the previously outlined Limnohabitans phylogeny (25). Kasalický et al. (7) proposed the division of the Limnohabitans genus into five lineages (LimA, LimB, LimC, LimD, and LimE), which is also supported by cell morphology. Furthermore, Limnohabitans microdiversity patterns across a large habitat set (161 habitats) have been proposed using specific reverse line blot hybridization (RLBH) probes, targeting as many as 18 subgroups within the genus (26).

Canyon-shaped reservoirs, built by damming the original river valleys, are often characterized by pronounced longitudinal heterogeneity reflected in various parameters, such as water temperature, nutrient supply, type of organic matter, phytoplankton biomass, and primary production (e.g., see references 27 and 28). The canyon-shaped Římov reservoir (Czech Republic) is a typical example of such an ecosystem. After having been impounded in the late 1970s, it has been the focus of numerous scientific investigations (currently over 100 scientific papers) concerning different limnological aspects (e.g., see references 13, 29, and 30).

In the present 2-year investigation, the Římov reservoir was used as a study site to investigate the microdiversity and spatiotemporal development of Limnohabitans lineages along its longitudinal axis (Fig. 1) and partly also in the epilimnetic vertical profile in two studies focusing specifically on (i) the spring plankton development (sampled 3 times per week) covering the phytoplankton peak formation and decline (clear-water phase) and (ii) the seasonal succession of Limnohabitans populations during 1 year (sampled approximately three times per month).

FIG 1.

Locations of the sampling stations along the longitudinal axis of the Římov reservoir. Sampling points: dam, dam Z_eu, planak, middle, and inflow.

We intended to gather information on several intriguing and unsolved questions dealing with Limnohabitans bacteria, such as the following. (i) Do different lineages of the genus undergo seasonal and spatial succession? (ii) To what extent do the groups reflect environmental gradients? (iii) Is there a relation between certain algal groups and certain subgroups of Limnohabitans bacteria? To answer such questions, we can advantageously exploit the natural heterogeneity inherent in canyon-shaped reservoirs in combination with their seasonal dynamics using RLBH lineage-specific probes (26). We aim to bring novel insights to the dynamics of particular lineages of Limnohabitans bacteria under in situ conditions with significantly enhanced taxonomic resolution, in contrast to older Limnohabitans studies, which used only the general R-BT065 fluorescence in situ hybridization (FISH) probe and commonly targeted 5 of the 6 lineages of the genus Limnohabitans (7).

RESULTS

Bacterial community characteristics in the Římov reservoir.

Bacterial abundance in the Římov reservoir ranges from 2 × 10⁶ to 8 × 10⁶ cells/ml, of which 5 to 18% represent RBT bacteria (Fig. 2). The bacterial abundance, relative contribution of the Limnohabitans RBT group (lineages LimBCDE), and the numbers of subtypes detected by RLBH did not substantially vary during either of the conducted campaigns (Fig. 3 and 4; see also Fig. S1 and S2 in the supplemental material). We detected the same subtypes in both studies, except for Lim13+ found only in 2009 and Lim6+A found only in 2011. Four more probes (Lim5+, Lim6+B, Lim12+, and Lim12+A) gave no signal during either campaign and hence these subtypes were most likely not present. Lim7+A was detected only once (6 September 2011, dam station, euphotic zone; data not shown) and with a very weak signal.

FIG 2.

Situation in the Římov reservoir at the surface and at the euphotic depth of the dam during spring 2009. Development of phytoplankton biomass (A), extracellular primary production (B), total bacterial abundance (C), proportions of RBT bacteria (D), numbers of Limnohabitans subtypes (reverse line blot hybridization groups) (E), and phytoplankton composition (F and G).

FIG 3.

Shifts in Limnohabitans subtype (RLBH probes) composition in the Římov reservoir at the dam (surface) during the spring phytoplankton bloom and clear-water phase in spring 2009. Development of phytoplankton and predators (Cladocera, rotifers, and heterotrophic nanoflagellates [HNF]) (top), numbers of subtypes (RLBH probe signals) (middle), and Limnohabitans subtype temporal composition (bottom).

FIG 4.

Shifts in Limnohabitans subtype composition in relation to phytoplankton development. An overview of total bacterial numbers, R-BT065 proportions, chlorophyll a concentrations, phytoplankton main composition (cryptophyte and cyanobacterial proportions of total chlorophyll a) and numbers and types of Limnohabitans subtype groups of the downstream stations (inflow, middle, and dam) of the Římov reservoir from April to November 2011. Data on phytoplankton composition at inflow are not available.

Spring temporal changes of distinct Limnohabitans subtypes.

During the detailed spring 2009 campaign, we did not find significant differences between the two sampled layers of the lacustrine part of the Římov reservoir (Fig. 2). The dam surface layer (dam) had slightly richer phytoplankton biomass, bacterial abundance, and RBT bacterial proportions than the euphotic depth (dam Z_eu). However, the largest difference was in extracellular primary production, which was one order of magnitude higher in dam samples than dam Z_eu samples. Despite these differences, the two layers did not substantially differ in RLBH probe detection, and the same Limnohabitans subtypes commonly occurred in both layers (Fig. 3; see also details in Fig. S1).

The number of Limnohabitans subtypes (Fig. 2E and 3) reflected the ongoing spring succession in the water. We observed a relatively stable period of the spring overturn with seven subtypes detected. During the phytoplankton spring bloom, the number of groups increased to nine, together with the intensity of the signal indicating a higher abundance of these subtypes. Bacterial abundance at the dam was the highest at the very beginning of the clear-water phase, but the number of subtypes decreased to the minimum of 5 at the end of the clear-water phase. It then started to rise again as phytoplankton biomass (represented mainly by diatoms and cryptophytes) increased (Fig. 2F and G).

Limnohabitans subtype composition in spring.

In spring, we detected altogether 12 RLBH-probe positive signals (subtypes). The detected subtypes did not occur at the same time, but they showed a clear trend (Fig. 3; see also details in Fig. S1). Lim1+, Lim2+, Lim4+, Lim3+, and Lim7+B were present all the time. Lim8+ was detected only in the early spring during the mixing phase and disappeared after stratification was established. Lim9+ appeared only for a short time at the end of the clear-water phase. Lim11+ emerged only at the end of the phytoplankton spring bloom and lasted during the clear-water phase until the end of the sampling period. Lim3+ started in the early spring and lasted until mid-April. Lim4+ had a trend similar to that of Lim3+, but reappeared at the end of the clear-water phase. Lim7+C and Lim7+D emerged with increasing phytoplankton biomass and lasted even during the clear-water phase, while Lim13+ also emerged (with a small delay) at the beginning of the phytoplankton spring peak and disappeared after its collapse. The cooccurrence of the Limnohabitans subtypes is summarized in Fig. S3.

Relationship to phytoplankton and zooplankton.

The phytoplankton at the dam and dam Z_eu sites in spring consisted mainly of small green flagellates (Chlamydomonas sp.), cryptophytes, chrysophytes, and diatoms (Fig. 2F and G). There were pronounced seasonal patterns in the abundance of heterotrophic flagellates, rotifers, and cladocerans (Fig. 3). The RLBH probe signal data (present/absent) were used to statistically evaluate the relationship between Limnohabitans subtype emergence and phytoplankton or zooplankton presence in the spring of 2009 (Fig. 5A). A redundancy analysis (RDA) showed a tight positive correlation of Lim7+C, Lim7+D, and Lim13+ with chlorophyll a, chrysophytes, cryptophytes, and heterotrophic nanoflagellates, which coincided with the spring algal bloom and bacterivory peaks (Fig. 3). Lim3+ and Lim4+ were correlated positively with ciliates and negatively with zooplankton, while Lim9+ and Lim11+ were positively correlated with zooplankton. Many subtypes completely disappeared during the peak cladoceran presence between 6 May and 16 May, with only a few subtypes persisting during this period.

FIG 5.

Redundancy analysis (RDA) of the occurrence of Limnohabitans probe-defined groups (subtypes) in relationships with phytoplankton and zooplankton in the Římov reservoir during phytoplankton spring peak in 2009 (63% of explained variability), pooled data from dam and dam Z_eu (A), and during the entire season in 2011 (25% of explained variability), pooled data from middle, dam, and dam Z_eu sites (B).

Relationship to water chemistry.

We observed a clear pattern in the relationships between Limnohabitans subtypes and water chemistry during the spring (see Fig. S4). A very tight positive correlation was found between the early spring subtype Lim8+ and soluble reactive phosphorus (SRP; F = 168.5; P < 0.001), while the same subtype was negatively correlated with temperature. Subtypes that occurred at the end of the phytoplankton peak and in the clear-water phase (Lim9+ and Lim11+) were correlated positively with temperature but negatively with total phosphorus and oxygen. The group of subtypes prevailing during the phytoplankton peak (Lim7+C, Lim7+D, and Lim13+) was correlated positively with chlorophyll a and pH but negatively with dissolved inorganic carbon (DIC) and SRP. The group of subtypes cooccurring mainly in the first half of the spring sampling (Lim3+ and Lim4+) was positively correlated with total phosphorus and oxygen. Temperature turned out to be unimportant for most of the groups, apart from the positive correlation with the Lim11+ group and negative correlation with Lim8+.

Spatiotemporal patterns along the longitudinal transect of Římov reservoir.

The Římov reservoir was sampled from April until November 2011 at four stations (Fig. 1). The bacterial abundance during this period was more or less stable (average of 4 × 10⁶ ml⁻¹), but the RBT proportions and the number of detected subtypes showed a rather decreasing trend (Fig. 4A, D, and G). The highest numbers of Limnohabitans subtypes were found at the inflow area on all sampling dates. While there were no differences in bacterial abundance from the inflow part downstream, the rest of the reservoir had mostly about one-half of the RBT bacteria and subtype numbers compared to the inflow station (Fig. 6). In April, the RBT bacteria comprised 10 to 18% of total bacteria with 5 to 9 subtypes detected, but in November, they comprised less than 5% of RBT bacteria with 3 subtypes only.

FIG 6.

Differences among sampling stations in bacterial abundance (A), relative proportions of R-BT065 (B), and abundance of Limnohabitans subtypes (C), April to November 2011. P and F values represent differences of inflow to the other sampling stations (analysis of variance [ANOVA]).

Limnohabitans subtypes during the season.

In general, we detected the same numbers of probe-defined Limnohabitans groups in the entire season of 2011 (spatial study with less frequent sampling) as during the dense 2009 spring campaign (Fig. 4; see also details in Fig. S2). Unlike in 2009, we detected two new subtypes Lim6+A and Lim7+A, from which Lim7+A was detected only once, and so it was excluded from further analyses. The subtype Lim13+, which persisted in the water for more than 1 month in 2009 (Fig. 3; see also Fig. S1), was completely missing. The inflow area turned out to be the richest in terms of the numbers of subtypes detected (Fig. 6). Other stations displayed rather similar patterns, with detection of Lim1+, Lim2+, Lim4+, Lim3+, Lim7+B, Lim7+C, and Lim7+D. Lim8+ was found solely in spring and Lim9+ in late fall. Although the inflow always had the highest subtype number, Lim6+A and Lim9+ were never observed there. On the other hand, lineages Lim3+ and Lim4+ were detected mainly at the inflow in 2011 and very scarcely in the rest of the reservoir.

Relationships between Limnohabitans subtypes and phytoplankton.

When analyzing the relationship between Limnohabitans subtypes and algal composition, we observed a positive correlation between cryptophytes and Lim7+C plus Lim7+D (Fig. 5B). In summer, a massive cyanobacterial bloom of Aphanizomenon flos-aquae appeared in the lacustrine parts of the reservoir, being the most prominent at the middle station (Fig. 4E). A tight positive correlation emerged for Lim6+A and cyanobacteria (Fig. 5B), whereas we discovered strong negative correlations between cyanobacteria and both the Lim1+ and Lim2+ subtypes, which were normally always present. Lim6+A, a completely new subtype, suddenly appeared and persisted until the bloom vanished (Fig. 4F). Lim11+ was another subtype that was detected concomitantly during the spring phytoplankton peak and summer cyanobacterial bloom at all stations for a longer period.

Overall significant trends in the data.

During our 2-year observations, we detected positive correlations between the proportions of RBT bacteria and the numbers of their subtypes (r² = 0.56; P < 0.001). There was a positive correlation between phytoplankton (both biomass and chlorophyll a) and RBT bacteria and the numbers of subtypes. However, this relationship strongly depends on which phytoplankton species are present. There was a significant increase (r² = 0.633; P < 0.001) in the number of subtypes during the spring phytoplankton peak and a significant decrease (r² = −0.582; P < 0.01) in the subtype numbers during the summer cyanobacterial bloom. Furthermore, the number of subtypes was significantly positively correlated with cryptophytes in both years, in 2009 (r² = 0.54; P < 0.01) and in 2011 (r² = 0.365; P < 0.01). Specifically, the subtypes Lim7+C, Lim7+D, and Lim13+ seem to be positively correlated with cryptophytes (Fig. 5). Lim11+ was positively correlated with zooplankton (Fig. 5A) and Lim 6+A with cyanobacteria (Fig. 5B). Limnohabitans lineages targeted by the Lim1+ and Lim2+ probes represented a generalist behavior (persisters) and showed strong signals almost throughout the entire study.

DISCUSSION

Current scientific methods allow us to go below the generic level of taxonomic resolution and to study the seasonal dynamics or distribution of various lineages within one genus. Searching for bacterial ecotypes, the phylogenetic placement of pyrosequences (targeting V1-V2 regions of the 16S rRNA gene) was successfully used to distinguish the dynamics of SAR11 subclades during 9 years (31). In addition, Hahn et al. (32) used sequencing-based methods to obtain the intrataxon biogeographic pattern of Polynucleobacter necessarius subspecies asymbioticus. Salcher et al. (33) tracked two planktonic freshwater lineages of Methylopumilus (Betaproteobacteria) for 4 years using specific catalyzed reporter deposition fluorescence in situ hybridization (CARD-FISH) probes. Earlier, Wu and Hahn (34) used the same method to follow and predict the appearance of the Polynucleobacter B subcluster in two consecutive years. However, species-specific probes can be difficult to design due to the lack of resolution on ribosomal genes.

In our study, we used the RLBH method that enabled an entirely new approach to reveal the dynamics of genotype-like (subtype) groups under natural conditions. Despite the semiquantitative nature of the results, this method revealed spatiotemporal trends in the distribution of distinct Limnohabitans subtypes.

Seasonality of Limnohabitans subtypes.

As already known from the literature, seasonality and annual patterns in bacterioplankton community variability can be highly inconsistent on a seasonal and annual scale and show little similarity from year to year (35). A clear seasonal succession of the bacterioplankton community was described in the comprehensive study of Eiler et al. (36). In the latter study, the temporal dynamics of previously designed tribes (3) revealed an extensive synchrony and associations with seasonal events, such as ice cover, ice-off, mixing, and phytoplankton dynamics.

Based on cautious comparisons between the 2009 spring monitoring and the spring part of the entire season study conducted in 2011, we may, to a certain extent, predict the development of some Limnohabitans lineages. Distinct Limnohabitans subtypes showed either a temporal restriction in their distribution (e.g., lineages targeted by the Lim8+ probe, detected during the spring phytoplankton peaks) or a rather broad, i.e., temporally or spatially unspecific, distribution (e.g., lineages targeted by the Lim1+ and Lim2+ probes). The subtypes Lim1+ and Lim2+ together form a monophyletic lineage called LimB, which seems to be the most abundant and persistent Limnohabitans group (7, 26). The abundance of this group suggests that it is quite unresponsive toward changes in environmental conditions. It also implies a rather broad definition and delineation of some of the probes and thus their taxonomic, and likely also ecological, “unspecificity” (26). However, we are aware that more detailed studies are needed to make well-based predictions.

Relationship to phytoplankton.

It was previously shown that phytoplankton affects the composition and dynamics of bacterial assemblages, caused probably by the nature and amount of released extracellular primary production (EPP) (23, 37–39). As suggested earlier (9, 38), Limnohabitans bacteria may be able to utilize algal exudates and other algal-derived substrates (22). They are opportunistic fast-growing bacteria that take advantage of the high phytoplankton densities in spring (10, 40, 41). In this study, we provided deeper insights into the possible relationships of selected Limnohabitans lineages to certain phytoplankton groups by showing significant correlations between the variables (Fig. 5A). In Šimek et al. (9), the most significant increase in the abundance of Limnohabitans strains was observed in experiments with Limnohabitans cultures growing together with Cryptomonas sp., which is in accordance with the results obtained in this study. Cryptophytes are known to produce a variety of exudates that the bacteria can readily take up and use. Here, we further specifically confirmed a positive effect of cryptophytes biomass on the Limnohabitans lineages targeted by the probes Lim7+C, Lim7+D, and Lim13+ (Fig. 5A). In Šimek et al. (9), strain II-D5 (Lim9+) exhibited much stronger growth in a mixed culture with Cryptomonas than strain II-B4 (Lim8+), indicating a more efficient utilization of alga-derived substrates produced by cryptophytes. A similar result can be seen in Fig. 5, where lineage Lim8+ is negatively correlated with cryptophytes. This negative correlation is a result of the time period when both species occurred. Lineage Lim8+ can be detected only in the early spring during the mixing phase when phytoplankton biomass, and especially cryptophyte biomass, is still very low, whereas Lim9+ is connected to warmer water with cryptophytes and the occurrence of Daphnia. Thus, there appears to be a partial niche separation of these two closely related bacterial species.

Experiments conducted in the Římov reservoir showed significant negative effects of a massive bloom of Microcystis aeruginosa on uptake rates of the RBT cluster as well as its population density (42). Notably, laboratory stock cultures of cyanobacteria also supported very limited cooccurring populations of Limnohabitans bacteria (R-BT065-probe targeted) compared to that of other phytoplankton groups (9). Similarly, as in Horňák et al. (42), we also observed a strong negative effect of a massive cyanobacterial bloom of Aphanizomenon flos-aquae on Limnohabitans (Fig. 4E and F). However, interestingly, the Lim6+A and Lim11+ lineages profited, as they were the only groups detected during the toxic cyanobacterial bloom. It is not known whether they used the new niche space created by the disappearance of other lineages that were unable to grow together with the toxic cyanobacteria or if they used some specific nutrients produced by the cyanobacteria (e.g., organic growth factors or even cyanotoxins).

Relationships to protists and zooplankton.

Bacterivorous protists can greatly impact emerging groups of bacteria, which is reflected by their ecological adaptations. Even though it has been proven (13, 43, 44) that the RBT bacteria are significantly selected for by bacterivorous flagellates compared to other bacterial lineages, no clear effect on the relative contribution of RBT to total bacteria was detected during the spring campaign at the dam site (Fig. 2C and D). However, our current data offer deeper insights into the factors which may modulate the population dynamics of the detected lineages. Interestingly, aside from the omnipresent Lim1+ and Lim2+ lineages, the Lim7+B and Lim7+C lineages showed strong signals, not only mainly over the spring phytoplankton peak but also through the entire period (approximately 14 April to 6 May), with the strongest grazing pressure by protists (Fig. 3; see also Šimek et al. [13]). It seems likely that the lineages Lim1+, Lim2+, Lim7+B, and Lim7+C were capable of compensating for large grazing losses (for details see Šimek et al. [13]) by increasing their growth rate. Thus, these represent grazing-resistant lineages. In contrast, lineages Lim3+ and Lim4+, which are examples of grazing-vulnerable lineages, rapidly decreased with the onset of the grazing peak, while the Lim4+ and Lim3+ lineages persisted until the middle of the period with high grazing pressure (see Fig. S1 in the supplemental material). These data further indicate that either some lineages were more vulnerable to grazing than others or their growth potential was simply insufficient to override the negative effect of the enhanced flagellate bacterivory.

Filtering zooplankton select for larger cells of all microbial groups, namely, algae, protists, and bacteria. During the clear-water phase, the mean cell volume of bacteria can dramatically decrease, and mainly small cells prevail in both absolute and relative numbers (13). Berga et al. (45) showed that the presence of Daphnia magna affected the community composition and reduced bacterial diversity but not bacterial abundance. In our case, bacterial abundance also did not change much, but the relative proportions of RBT bacteria (Fig. 2D) and the detected subtype numbers (Fig. 2E) remarkably decreased, which was probably caused by a generally larger cell volume of Limnohabitans and the domination by daphnids with fine filter mesh sizes. Notably, the omnipresent subtypes (Lim1+ and Lim2+), represented by small cells (cf. Table 1 and Kasalický et al. [7]), survived the heavy filtration of Daphnia spp.

TABLE 1.

Group-specific probes used in the Limnohabitans RLBH assays, their 5′ to 3′ sequences, and melting temperatures^a

Probe name	Sequence	Tm (°C)	Affiliation	Targeted strain(s)
Lim1+	CTGTGTCAAAGAGTTATTCACATT	55.9	LimB	Rim11, KL1, Hippo4
Lim2+	AAACTTTGTTCGCATTACGGC	55.9	LimB	2-KL15, clone-17, clone-72, DAM0.8-3
Lim3+	AAATAGCTTTGATCTTGAAAGAGGT	56.4	LimA/LimE	B10-3v, SP3
Lim4+3+	ATTGATTGATTAACTAGGCTGTTC	55.9	LimA/LimE	L. curvus, L. australis, Rim8, Jir61, B10-3v, SP3
Lim4+	AGATATCAGAGTTRCTAGCGG	56.9	LimA	L. curvus, L. australis, Rim8, Jir61
Lim5+	CGGCTGAGGCGTAAGC	56.9	LimC (C4)	2KL-17, Bal53
Lim6+A	ACGACTTGTGCGCATGCT	56.0	LimC	Only clones
Lim6+B	CGCGTAAATCGAATAAATCCAATA	55.9	LimC (C6)	2KL-3, 2KL-7, WS1, SP2, G3-3
Lim7+A	CGCAAGCCTCGAGTCATT	56.0	LimC	2KL-1
Lim7+B	CGCAAGCCCAAGTCATTG	56.0	LimC (C4)	B22-3k, Hin4
Lim7+C	CTTATCAAAGGTTTTGATCTCATTC	56.4	LimC (C4)	15K
Lim7+D	ACTTATCAAATGTTTTGATCTCATTCAA	56.3	LimC (C4)	Jir75
Lim8+	TATCGAGTGTTAATRGTGTCTGA	56.2	LimC (C2 and C3)	L. parvus II-B4, LI2-55, T6-20, T6-5
Lim9+	GGGCCTTGCAGTGGC	56.0	LimC (C1)	L. planktonicus II-D5, 2KL-16, Rim42
Lim10+	TTGAGCGGATCCTGCAAG	56.0	LimC	KL6
Lim11+	AAGAGATTGCGAGGCTGTTTT	55.9	LimC (C5)	2KL-27, KL5
Lim12+A	GTTCCCGTAAGGGACTTTAT	55.3	LimC (C5)	Only clones
Lim13+	GGGTCTTGCAAGGGCC	56.9	LimC	Mo2-6, 2KL-5

^aProbes target different regions of ITS 16S to 23S rRNA of the Limnohabitans genus bacteria. Adapted from reference 26.

Surprisingly, we detected only occasionally the subtype group Lim9+ (targeting Limnohabitans planktonicus strain II-D5^T). Moreover, Lim9+ was never detected in our previous study across 161 habitats (26). L. planktonicus was previously described as a competitive strain (21) able to maintain the bacterial population diversity in a grazer-free system (46). On the other hand, L. planktonicus was close to extinction in experiments with flagellate grazing (42, 44). Thus, the absence of Lim9+ in most samples might reflect the lack of a successful defense strategy, except a specific adaption to the epibiotic lifestyle of Daphnia (14) with which a beneficial symbiotic relationship (with Daphnia magna) was demonstrated previously (47–49).

Longitudinal effects of the reservoir.

Water circulation has a strong effect on nutrient dynamics (50) and microbial community patterns along the reservoir (51). In the entire season study, the highest abundance of the RBT cluster and diversity of Limnohabitans lineages were consistently found at the inflow, the only riverine station above the plunge point (Fig. 1). It seems that nutrients were the main driving force for such high Limnohabitans diversity, since algal biomass and production in the river inflow were rather low. This might be somewhat in contradiction with the generally accepted assumption of increasing diversity with increasing productivity (e.g., see references 50 and 52). However, it has to be noted that the inflow sampling station is located just upstream of the plunge point, which clearly separates the inflow and transient zones of the reservoir. The inflow parts of the reservoir have been suggested to be dynamic and prone to sudden disturbances brought about by weather, e.g., causing increased or decreased flowthrough, and also influenced by upstream-situated ponds and reservoirs, human activities in the watershed, and the terrestrial environment. In the case of the underflow and interflow (see reference 50), the plunge point represents a barrier of two water bodies. Two LimC subtype groups were never observed in the inflow, namely, Lim6+A, which emerged during the cyanobacterial bloom, and Lim9+, with a symbiotic relation to Daphnia (14), both typical for stratified waters. On the other hand, lineages Lim3+ and Lim4+ (both LimA) were detected mainly at the inflow station during 2011 and very scarcely in the rest of the reservoir. This finding can be explained by the LimA link to humic organic matter coming with the river (53). In some cases, seasonal shifts in the bacterioplankton community composition could be related to shifts in the source (terrestrial versus phytoplankton) and instability of dissolved organic matter (DOM) (54, 55). This was, to a certain extent, confirmed in our study, since the biodegradable dissolved organic carbon (DOC) concentration is frequently highest at the inflow, corresponding to the highest Limnohabitans diversity. On the other hand, several studies on the effects of DOC concentration on bacterial communities showed a tight relationship between these two parameters (56, 57), suggesting rather a negative effect of bulk DOC on the taxon richness of bacterioplankton communities. In the study by Fujii et al. (56), DOC was identified as one of the major factors affecting bacterioplankton community dynamics. In our study, we found no measurable effects of DOC on any of the Limnohabitans lineages (data not shown). However, dissolved inorganic carbon (DIC) showed a negative correlation with Limnohabitans lineages targeted by the Lim7+B, Lim7+C, Lim7+D, and Lim13+ probes and a positive relationship with Lim8+ (see Fig. S4). Similarly, DOC type strongly contributes to niche separation of major Polynucleobacter lineages (P. necessarius subspecies asymbioticus and the B subcluster [32]), but also between P. necessarius subspecies asymbioticus and Limnohabitans bacteria (12).

Nutrient impact.

There is a large gradient from the inflow to the dam station. Both station physical and chemical parameters markedly differ as demonstrated in data from 2007 (58) measured at the inflow and dam: average water temperature, 14.1°C and 18.8°C, respectively; total phosphorus, 86.9 μg/liter and 20.3 μg/liter, respectively; and soluble reactive phosphorus, 31.4 μg/liter and 3.5 μg/liter, respectively. Unfortunately, one cannot detect clear relationships between Limnohabitans subtypes and chemical data during the entire season that would allow for any firm conclusion. There was a weak effect of nutrients (mainly dissolved reactive phosphorus [DRP]) on all Limnohabitans lineages, which, to a certain extent, corresponds to a previous finding of high growth rates of RBT bacteria even under very low DRP concentrations (18). In that earlier study, RBT bacteria had the highest specific growth rates of all investigated bacterial groups. However, the RBT probe was used in that study, whereas we achieved a much finer coverage and resolution by using RLBH probes, but still, neither a positive nor negative effect of DRP on any of the Limnohabitans lineages was confirmed. This suggests generally high affinities of Limnohabitans lineages to DRP resources, thus indicating the rather minor importance of this parameter in modulating the population dynamics of all Limnohabitans lineages described so far. Similarly, generally no obvious effects of phosphorus species on the overall changes in bacterial composition were observed in eutrophic Lake Mendota (25). Interestingly, their results suggest that a particular bacterial group is not universally favored by increased nutrient loads to a lake. Therefore, efforts to predict which bacteria are involved in nutrient cycling during these periods must take into account the seasonality of the freshwater bacterial communities.

Concluding remarks.

Spatiotemporal dynamics of individual bacterial ecotypes are a key to understanding the bacterial community composition in marine or freshwater environments (33). Studies on the most abundant marine bacterium provided evidence for three SAR11 ecotypes, namely, spring, summer, and deep water (59). A deeper phylogenetic resolution strongly supported conclusions that seasonal factors are the main drivers for the subclade niche partitioning within this important oligotroph (31). On the other hand, the copiotroph Limnohabitans bacterial lineages seem to develop a dependence on other microorganisms (primary producers or grazers) (Fig. 5). The ecotype diversification includes important genome modifications (subtle for SAR11 [60]) leading to the changes in their ecological behavior. Regarding the comparable importance of SAR11 for the ocean (61) and Limnohabitans for freshwater (13), the particular ecotype presence may largely influence the entire microbial food web.

MATERIALS AND METHODS

Sampling, basic limnological parameters, and bacterial and Limnohabitans abundances.

Two sampling campaigns were conducted in the canyon-shaped, mesoeutrophic Římov reservoir (48°50′56″N, 14°29′26″E, 470 m above sea level, length, 13 km; mean volume, 34.5 × 10⁶ m³; area, 2.06 km²; mean retention time, 100 days) located in the south of the Czech Republic. More information on the reservoir can be found in Jezbera et al. (62).

The first campaign, performed in 2009, intensively covered the spring phytoplankton peak formation and decomposition; for details, see Šimek et al. (13). Samples were collected from the dam site (Fig. 1) at the surface (0.5 m) and the euphotic depth (dam Z_eu), which is the depth characterized by 1% of the surface light intensity. In total, we investigated 46 samples by RLBH for the occurrence of distinct probe-defined Limnohabitans lineages.

The second campaign focused on the downstream succession of Limnohabitans lineages along the longitudinal axis of the reservoir. It covered almost the entire vegetation season of 2011, starting from 20 April and ending 22 November, at 1- to 3-week intervals, yielding a total of 20 sampling dates. Samples were retrieved from 4 stations, namely, inflow, middle, planak, and dam (Fig. 1), plus the dam Z_eu layer (see above) to investigate the influence of light. A total of 100 samples were generated and further analyzed.

Assessments of basic limnological parameters, such as temperature, oxygen, pH, and conductivity, were performed in situ (WTW 330i oximeter, pH meter; WTW, Weilheim, Germany). Water samples for analyses of nutrients, microbial community composition, and genomic DNA isolation were taken from the sampled depths by a Friedinger sampler and instantly transferred to the laboratory and further processed.

Water samples for microbial community composition analyses were fixed with formaldehyde (for bacterial abundance) or paraformaldehyde (for CARD-FISH analyses)—both at a 2% final concentration—or they were filtered through 0.2-μm Poretics filters for subsequent DNA extraction (63, 64). Bacteria were enumerated using an Olympus AX70 epifluorescence microscope after DAPI (4′,6-diamidino-2-phenylindole) staining as described in Šimek et al. (16). Percentages of the RBT cluster of Limnohabitans bacteria were estimated using CARD-FISH (12, 65), employing the R-BT065 fluorescence probe (16).

Limnohabitans-specific PCR amplification of ITS sequences and reverse line blot hybridization assays.

Limnohabitans-specific amplification of 16S to 23S internal transcribed spacer (ITS) sequences from the DNA extracted from the Římov biomass samples was performed using primers Lim379F and LimCurvITS-R (biotinylated), as described in Jezbera et al. (26). The length of the PCR product was approximately 1,900 bp. Nested PCRs were performed in the cases with small amounts of PCR product (<50 ng · μl⁻¹) from the reservoir samples (26).

The subsequent reverse line blot hybridization (RLBH) assays, using biotinylated PCR products, were performed according to Jezbera et al. (26, 64). In total, 18 Limnohabitans-lineage specific RLBH probes were used (Table 1) covering a large percentage (92%) of currently available isolates (7). Probe results were scored as negative (0), weak (1), normal (2), or strong (3), corresponding to the intensity of the probe signal. They are further presented in a grayscale range in the overview of the data (Fig. S1 and S2 in the supplemental material). The semiquantitative approach was tested for each subtype using five dilutions of the positive control (1, 0.5, 0.1, 0.05, and 0.01). The method was fine-tuned to enable the use of semiquantitative data in our analyses. However, for the statistics and for the Limnohabitans-subtype behavior discussion, only presence/absence data were used.

Other chemical and biological variables.

The following variables were measured as previously detailed in Znachor et al. (51): phytoplankton biovolume (mm³ · liter⁻¹), chlorophyll a (μg · liter⁻¹), particulate phosphorus (μg · liter⁻¹), SRP (μg · liter⁻¹), and total phosphorus (TP; μg⁻¹ · day⁻¹). Extracellular primary production (EPP; μg · liter⁻¹ · day⁻¹), percentage of total primary production (%EPP), and dissolved inorganic carbon (DIC; μg · liter⁻¹) were measured as described in Šimek et al. (38). Phytoplankton were enumerated on an inverted microscope at 100 to 1,000× magnification (Olympus IX 71) after sedimentation of a known volume of sample in a counting chamber (66). The mean algal cell dimensions were obtained for biovolume calculation using the approximation of cell morphology to regular geometric shapes (67). The detailed description of phytoplankton composition at three sampling sites (dam, dam Z_eu, and middle) for the entire season 2011 can be found in Znachor et al. (51).

Statistical analysis.

Multivariate analyses were conducted using CANOCO (68). Principal-component analysis (PCA) and redundancy analysis (RDA) were carried out with centering and standardization by species normalization. Forward selection was used to choose significant explanatory (environmental) variables. Variables were included when a P value of < 0.05 was estimated by a Monte Carlo permutation test with 999 unrestricted permutations. The results of the analyses were visualized by CanoDraw for Windows (68). In the CANOCO graph, arrows going in the same directions represent positive correlations, while arrows in the opposite directions show negative correlations of the tested factors.