Lake Kivu is a meromictic lake located in the volcanic region between Rwanda and the Democratic Republic of the Congo and is the smallest of the African Great Rift Lakes. The monimolimnion of the lake contains a large amount of dissolved CO
2 and methane (300 km
3 and 60 km
3, respectively) as a result of geological and biological activity (
24,
73,
85). This massive accumulation converts Lake Kivu into one of the largest methane reservoirs in the world and into a unique ecosystem for geomicrobiologists interested in the methane cycle and in risk assessment and management (
34,
71,
72,
85). Comprehensive studies on the diversity and activity of planktonic populations of both large and small eukaryotes and their trophic interplay operating in the epilimnetic waters of the lake are available (
33,
39,
49). Recent surveys have also provided a deeper insight into the seasonal variations of photosynthetic and heterotrophic picoplankton (
67,
68), although very few data exist on the composition, diversity, and spatial distribution of bacterial and archaeal communities. In this regard, the studies conducted so far of the bacterial/archaeal ecology in Lake Kivu have been mostly focused on the implications on the methane cycle (
34,
73), but none have addressed the presence and distribution of additional archaeal populations in the lake.
During the last few years, microbial ecology studies carried out in a wide variety of habitats have provided compelling evidence of the ubiquity and abundance of mesophilic archaea (
4,
10,
13,
19). Moreover, the discovery of genes encoding enzymes related to nitrification and denitrification in archaeal metagenomes from soil and marine waters (
29,
86,
88) and the isolation of the first autotrophic archaeal nitrifier (
40) demonstrated that some archaeal groups actively participate in the carbon and nitrogen cycles (
56,
64,
69). In relation to aquatic environments, genetic markers of ammonia-oxidizing archaea (AOA) of the marine
Crenarchaeota group 1.1a (MCG1) have consistently been found in water masses of several oceanic regions (
6,
14,
17,
26,
28,
30,
37,
42,
51,
52,
89), estuaries (
5,
9,
26,
53), coastal aquifers (
26,
66), and stratified marine basins (
15,
41,
44). Although less information is available for freshwater habitats, recent studies carried out in oligotrophic high-mountain and arctic lakes showed an important contribution of AOA in both the planktonic and the neustonic microbial assemblages (
4,
61,
89).
The oligotrophic nature of Lake Kivu and the presence of a well-defined redoxcline may provide an optimal niche for the development of autotrophic AOA populations. Unfortunately, no studies of the involvement of microbial planktonic populations in cycling nitrogen in the lake exist, and only data on the distribution of dissolved inorganic nitrogen species in relation to phytoplankton ecology (
67,
68) and nutrient loading are available (
54,
58). Our goals here were to ascertain whether or not archaeal populations other than methane-related lineages were relevant components of the planktonic microbial community and to determine whether the redox gradient imposed by the oxic-anoxic interphase acts as a threshold for their vertical distribution in epipelagic waters (0- to 100-m depth). To further explore the presence and potential activity of nitrifying archaeal populations in Lake Kivu, samples were analyzed for the abundance and vertical distribution of signature genes for these microorganisms, i.e., the 16S rRNA of MCG1 and the ammonia monooxygenase subunit A (
amoA) gene by quantitative PCR.
DISCUSSION
The contribution of archaea to planktonic microbial assemblages in stratified freshwater lakes is variable, but reported values are usually lower than those measured for marine environments (
10). Although obtained by a different methodological approach, archaeal abundance in neighboring Lake Victoria reached 5.9% of the total nucleic acids (
38). Similar values (between 1 and 7% of DAPI-stained cells) were obtained by FISH and CARD-FISH in different freshwater lakes (
36,
59), although recent studies carried out in high mountain lakes reported abundances of up to 22% in Crater Lake (
87) and 37% in Lake Llebreta (
4). In stratified marine environments such as the Cariaco Basin and the Black Sea, the archaeal planktonic fraction ranged from 1% to 9% and from 10% to 30% of total DAPI counts, respectively, showing maximal abundances at the redoxcline coinciding with depth maxima of nitrite and nitrate (
15,
41,
44). Similar distribution patterns have also been reported by Pouliot and coworkers (
61) in two meromictic arctic lakes. Based on this prior research, the distribution and relative abundance (0.3% to 4.5% of total DAPI counts) of planktonic archaea in Lake Kivu agree with data available for other freshwater environments. It should be noted, however, that extensive studies are needed to ascertain if variations in environmental conditions between the rainy and dry seasons may affect the distribution and abundance of the planktonic archaeal community, as has been described for other microbial populations thriving in Lake Kivu (
23,
67,
68).
The phylogenetic structure of the archaeal assemblage in Lake Kivu was fairly homogeneous in all sampling basins, with a clear phylogenetic segregation imposed by the oxic-anoxic transition (see Fig. S3 in the supplemental material). Sequences from the anoxic water compartment mainly affiliated with the highly diverse miscellaneous crenarchaeotic group (MCG) (
32) and with methanogenic lineages. The MCG archaea are considered cosmopolitan (
83) but are frequently found in anoxic habitats such as deep subsurface marine sediments (
7) and hypolimnetic waters of sulfurous mesotrophic lakes (
45). Current evidence suggests that some members of the MCG lineage may obtain energy from the anaerobic oxidation of methane (
7,
83), a hypothesis that fits with the prevalent physicochemical conditions in the monimolimnion of Lake Kivu. OTUs assigned to methanogenic lineages grouped with either acetoclastic (identity values of 97.4% and 95.8% with
Methanosaeta concilii [OTU-1 and OTU-2, respectively] and uncultured
Methanosarcinales [OTU-3 and OUT-4]) or hydrogenotrophic (95.9% identity of OTU-5 to
Methanocellula paludicola) representatives, agreeing both with the biological origin of methane in the lake and with the methanogenic archaeal groups commonly found in other stratified lakes (
43). The recovery of a few sequences related to methanogens from oxygenated water layers (bands
aK3,
aI3, and
aB1 in Fig. S1 in the supplemental material) is, however, not in accordance with the strictly anaerobic metabolism assumed for these microorganisms. The oxygen tolerance of some members of the
Methanosaeta cluster (
31) or the occurrence of water-mixing processes that transported microorganisms from the upper part of the monimolimnion to shallow depths might explain these findings.
At the oxic-anoxic interphase and above, almost all the recovered sequences grouped within archaeal lineages containing ammonia-oxidizing representatives, i.e.,
Crenarchaeota group 1.1a and group 1.1b (
64). Whereas all sequences affiliated to the former were assigned to a single OTU (OTU-8) related to the nitrifying marine archaeon “
Candidatus Nitrosopumilus maritimus” (Fig.
4), those related to group 1.1b were distributed in 11 OTUs. Since this lineage is mainly composed of crenarchaeal phylotypes recovered from soil (
8,
57), the high level of richness found in Lake Kivu for group 1.1b raises the question of whether these phylotypes were indigenous from the plankton or were introduced to the lake by surface runoff. The fact that most of the group 1.1b-related sequences were recovered from the southern basin and particularly Bukavu Bay, which are basins partly isolated from the main lake by sills of different depths (
18,
81) and receive high water inflows by rivers and subaquatic sources (
18,
54,
72), points to a terrestrial origin of the detected phylotypes. On the other hand, the close affiliation of some of the OTUs within
Crenarchaeota group 1.1b with archaeal sequences potentially involved in ammonia oxidation in soils (e.g., 98.8% identity of OTU-16 to fosmid soil clone 54d9 [
86]) is relevant. In any case, all archaeal
amoA sequences recovered from Lake Kivu composed a homogeneous subcluster within the freshwater clade (Fig.
6) (
26) and clearly separated from marine and terrestrial sequences. This result suggests either that the phylotypes assigned to group 1.1b recovered from Lake Kivu are actually not able to oxidize ammonia or that the
amoA primers used (see Table
1) (
26) present some bias toward marine
amoA sequences. The use of different primer pairs for
amoA fingerprinting (
26) and qPCR (
89) and the different sensitivity of both techniques toward less abundant phylotypes hinder the proper comparison of the data. Thus, further investigation is needed to resolve the actual nitrification capacity of the soil-related phylotypes found in the water column of Lake Kivu, especially considering that the most frequent phylotypes retrieved affiliated with the marine clade.
The increase in archaeal cell numbers at the oxycline (30 to 50 m depth) and the concomitant vertical distribution of molecular signatures of marine ammonia-oxidizing crenarchaeota (MCG1 16S rRNA and
amoA genes) and nitrate and nitrite maxima at these depths agree with results found in other aquatic environments (
15,
41,
42,
52). Unfortunately, logistic problems during field sampling did not permit the preservation of the samples in such a way to allow further analysis of
amoA transcripts. Caution must therefore be exercised when considering the potential role of nitrifying crenarchaeota in Lake Kivu. Recent studies demonstrate that “
Candidatus Nitrosopumilus maritimus” strain SCM1 is adapted to extreme nutrient limitation (
50). According to these authors,
Nitrosopumilus-like AOA may benefit from this adaptation to compete for nitrogen sources with ammonia-oxidizing bacteria, heterotrophic bacterioplankton, and phytoplankton. Although it is far from being resolved whether all the marine AOA are specialized oligophiles like strain SCM1, the low ammonia concentrations found in the epilimnetic waters of Lake Kivu (<0.1 μM) (
58) may provide an optimal niche for their growth. In this regard, the low level of diversity of archaeal phylotypes in the oxycline and the low level of richness of
amoA genes found suggest that freshwater members of the
Crenarchaeota group 1.1a compose a distinct population at these depths. The identity values between the phylotype found in Lake Kivu samples and the reference strain “
Candidatus Nitrosopumilus maritimus” (94.9% to 96.2%) point to a weak phylogenetic relation probably linked to its freshwater origin. The homogeneous clustering of
amoA sequences recovered from Lake Kivu samples into the same freshwater clade provides further support to this hypothesis and to the idea of sequence clustering according to habitat (
6,
26,
61). In this regard, several authors have recently shown that salinity is a major driver affecting archaeal distribution either at a local or at a global scale (
4,
47,
53).
As stated above, the actual role of nitrifying crenarchaeota in the nitrogen cycle of Lake Kivu is, however, far from being resolved. Further activity and expression measurements are needed to confirm the significance of autotrophic archaeal nitrification in the lake and to determine the specific contribution of
Crenarchaeota group 1.1a and 1.1b in comparison to bacterial nitrifiers. This topic is of special interest, especially considering the small contribution of archaeal cells in the total microbial planktonic assemblage measured in this work. Although recent reports on the deep marine subsurface biosphere highlighted that some important microbial activities can be performed by a small, but very active, subset of community members (
27), further studies covering spatial and temporal variations of AOA populations in Lake Kivu should be addressed to ascertain their dynamics and seasonal abundance. Finally, the oligotrophic nature of neighboring lakes Tanganyika and Malawi offers potential habitats for the development of AOA. The finding of crenarchaeotal membrane lipids in sediments from these and other African lakes (
62,
63,
74,
84) supports this assumption. Further surveys focused on these aspects will provide a better picture of the processes and players beneath the microbial cycling of nitrogen in large oligotrophic African lakes.