|Home | About | Journals | Submit | Contact Us | Français|
In addition to the benthic and pelagic habitats, the epiphytic compartment of submerged macrophytes in shallow freshwater lakes offers a niche to bacterial ammonia-oxidizing communities. However, the diversity, numbers, and activity of epiphytic ammonia-oxidizing bacteria have long been overlooked. In the present study, we analyzed quantitatively the epiphytic communities of three shallow lakes by a potential nitrification assay and by quantitative PCR of 16S rRNA genes. On the basis of the m2 of the lake surface, the gene copy numbers of epiphytic ammonia oxidizers were not significantly different from those in the benthic and pelagic compartments. The potential ammonia-oxidizing activities measured in the epiphytic compartment were also not significantly different from the activities determined in the benthic compartment. No potential ammonia-oxidizing activities were observed in the pelagic compartment. No activity was detected in the epiphyton of Chara aspera, the dominant submerged macrophyte in Lake Nuldernauw in The Netherlands. The presence of ammonia-oxidizing bacterial cells in the epiphyton of Potamogeton pectinatus was also demonstrated by fluorescent in situ hybridization microscopy images. By comparing the community composition as assessed by the 16S rRNA gene PCR-denaturing gradient gel electrophoresis approach, it was concluded that the epiphytic ammonia-oxidizing communities consisted of cells that were also present in the benthic and pelagic compartments. Of the environmental parameters examined, only the water retention time, the Kjeldahl nitrogen content, and the total phosphorus content correlated with potential ammonia-oxidizing activities. None of these parameters correlated with the numbers of gene copies related to ammonia-oxidizing betaproteobacteria.
In ammonium-rich environments such as eutrophic lakes, ammonia-oxidizing Betaproteobacteria (β-AOB) perform the first, often rate-limiting step in the process of nitrification, hence playing an important role in the nitrogen turnover in a wide range of natural and artificial habitats (31). Their monophyletic nature allowed the successful application of molecular techniques based on the genes coding for the 16S rRNA gene and the A subunit of the ammonia monooxygenase enzyme (amoA). β-AOB have been considered an ideal model group in molecular microbial ecology (31). The last few decades have seen significantly increased numbers of studies focusing on diversity (7, 22, 24, 44, 48, 54) and niche differentiation and related driving factors (8, 13, 14, 30, 33, 45), as well as on the abilities of ammonia-oxidizing bacteria to cope with contaminants (40, 49, 53), environmental stresses (18, 34, 39), and global change (23, 43). The discovery of the process of anaerobic ammonia oxidization (42), together with the isolation of members of the kingdom Crenarchaeota able to oxidize ammonia (26), is currently changing and deepening the knowledge and understanding of the microorganisms involved in the nitrogen cycle.
Nitrification in freshwater as well as in shallow marine lagoon systems has been assumed to be associated with the sediment rather than with the overlying water (2). However, when planktonic nitrification rates are integrated over the whole water column, Vincent and Downes (59) demonstrated the impact of the pelagic community on the total nitrification process in lakes. In shallow freshwater lakes populated by large stands of macrophytes, the role of epiphytic nitrification must also be taken into account, since submerged macrophytes can provide a large accessible surface area for attached microorganisms (61). The bacterial ammonia oxidizers inhabiting the epiphytic compartment have been the subject of a limited number of studies. Eriksson and colleagues (15-17) measured the nitrification rates on the leaves and litter of submerged macrophytes, and Körner (29) reported that considerable numbers of ammonia-oxidizing bacteria colonize the leaves of different species of submerged macrophytes by means of most-probable-numbers counts. In a recent study of the β-AOB in shallow freshwater lakes (12), we described in a qualitative way that the epiphytic communities are composed of members of the Nitrosomonas oligotropha lineage and cluster 3 of the Nitrosospira lineage (48).
The present study focused on a more quantitative estimation of the ammonia-oxidizing bacteria in the epiphyton of two different submerged macrophytes, i.e., Potamogeton pectinatus and Chara aspera, present in three shallow freshwater lakes, which form part of the series of lakes studied before (12). We hypothesized that the numbers of cells would differ between the epiphytic, benthic, and pelagic compartments, as previously observed for the community composition, due to the prevailing environmental conditions (12).
In July 2005, samples of sediment, water, and submerged macrophyte leaves were collected from three shallow freshwater lakes (i.e., Lakes Gooimeer, Nuldernauw, and Vossemeer) in the center of The Netherlands. The lakes belong to a complex hydrological system of seven interconnected artificial shallow freshwater lakes created from the reclamation of 1,490 km2 of brackish area and described in more detail by Coci et al. (12). Restoration measurements have led to the reestablishment of submerged macrophytes in these shallow lakes, which enabled us to study these lakes for the presence of epiphytic ammonia-oxidizing bacteria. The whole system of interconnected lakes is divided into three hydrological subunits by sluices, and one lake in each of these subunits was selected for study. Lake Gooimeer is a freshwater body of 2,757 ha, of which 34% is less than 1.5 m deep. It has mainly clayish sediment and a water retention time of approximately 150 days. The dominant submerged macrophyte species in this lake is Potamogeton pectinatus (50). Lake Nuldernauw amounts to 600 ha, of which 56% is less than 1.5 m deep. It has mainly sandy sediment and a water retention time of 45 days. Lake Nuldernauw has been subjected to stringent restoration procedures, which led to the establishment of dense charophyte meadows and, subsequently, higher Secchi depth readings and a lower level of resuspension of sediment particles. Lake Vossemeer is the smallest lake (400 ha) and is also the most shallow of the lakes sampled (61% of the lake is less than 0.5 m deep). It has a sandy sediment and a water retention time of only 3 days. As in Lake Gooimeer, the dominant submerged macrophyte species in Lake Vossemeer is P. pectinatus (50).
On the basis of the distribution and the coverage percentages of the dominant submerged macrophyte species, i.e., P. pectinatus or C. aspera (50), three stations (stations A, B, and C) were selected in each of the lakes and were used as replicates for different analyses. The locations of the stations differ in their exposures to the dominant southwestern wind and their distances to the sheltering southern shores. The main physical and chemical characteristics of the sampling stations are summarized in Tables Tables11 to to3.3. Data related to the water retention time and the oxygen, Kjeldhal nitrogen, and phosphate concentrations in the water of the lakes were provided by the Rijkswaterstaat Directory IJsselmeergebied (RDIJ). Analytical analyses of sediment samples were performed as described by Coci et al. (12).
At every station, five replicate areas were randomly chosen by means of a 1-m2 polyvinyl square floating device. In each area, water samples were collected in triplicate by means of plastic bottles; 250 ml water was filtered on board over 0.2-μm-pore-size membrane filters for DNA isolation and over glass fiber filters for activity measurement. The leaves of submerged macrophytes were also collected in triplicate and were stored in plastic bags containing water from the station. Sediment samples were collected in five replicates by taking the upper 5 cm of two sediment cores within the 1-m2 sampling area.
To estimate the numbers of active ammonia-oxidizing cells in the epiphytic, benthic, and pelagic compartments, potential ammonia-oxidizing activities (PAAs) were determined in 50 ml of mineral medium with NH4+ at a final concentration of 1 mM, according to the protocol of Belser and Mays (3), as modified by Verhagen and Laanbroek (57). Briefly, the linear production of nitrite plus nitrate over time under optimal substrate, pH, and temperature conditions is taken as a measure of the potential rate of ammonium oxidation. Potential activities were expressed per m2 to be able to compare the different compartments mutually. Hence, for the pelagic zone, the activity per ml was multiplied by the depth of the water column at the sampling station to calculate the activity per m2. For the epiphytic zone, the activity per g (fresh weight) was multiplied by the amount of macrophyte biomass per m2. It was assumed that the submerged macrophyte plant biomasses at the sampling stations were, on average, 500, 100, and 80 g/m2 for Lake Gooimeer, Lake Nuldernauw, and Lake Vossemeer, respectively (50). The smaller amounts at the last two lakes are explained by the difference in macrophyte species (P. pectinatus in Lake Gooimeer and C. aspera in Lake Nuldernauw) and by the lower deep-water depth at Lake Vossemeer than at Lake Gooimeer. Finally, the activity per m2 sediment was calculated by converting the weight of the upper 5 cm into volume by using its bulk density.
Environmental DNA was extracted from 0.5 g freeze-dried sediment, 125 ml lake water filtered over a 0.2-μm-pore-size membrane filter, or 0.1 g freeze-dried macrophyte leaves for triplicate samples per station. The samples were homogenized by vortexing with 1 ml cetyltrimethylammonium bromide (CTAB) buffer (63) and 0.5 g sterilized zirconia-silica beads (diameter, 0.1 mm) and were subsequently subjected to disruption by bead beating at a 5.0-m/s rotation for 60 s. After the addition of 5 μl proteinase K (20 mg/ml), the samples were homogenized, incubated for 30 min at 37°C, and then quickly vortexed after 15 min. After the addition of 150 μl of a 20% SDS solution, the samples were incubated for 1 h at 65°C in a Thermoblock apparatus and quickly vortexed every 15 to 20 min. After centrifugation at 10,000 × g for 10 min, 600 μl of supernatant was collected in 2-ml clean screw-cap tubes. The rest of the sample was reextracted with 450 μl CTAB buffer and 50 μl of 20% SDS solution, vortexed for 10 s, incubated for 10 min at 65°C, and centrifuged at 6,000 × g for 10 min. Again, 600 μl was collected, added to the previously extracted supernatant, mixed with 1 ml phenol-chloroform-isoamyl alcohol solution (25:24:1, vol/vol/vol), and centrifuged at 6,000 × g for 10 min. One milliliter of supernatant was collected and placed into a new screw-cup tube containing 700 μl isopropanol, and the tube was incubated for 1 h at 24°C. After 20 min of centrifugation at 15,000 × g, the isopropanol was decanted and the pellet was resuspended and washed with 1 ml 20% cold ethanol. This was followed by 5 min of centrifugation at 15,000 × g, decantation of the ethanol, drying of the pellet under vacuum centrifugation, and finally, resuspension in 100 μl water (Sigma). All samples were purified and concentrated in a final volume of 50 μl with an AMPure PCR purification system, according to the manufacturer's instruction (Agencourt Bioscience Corporation, Beverly, MA). Quantification was done with 2-μl DNA samples and an ND-1000 apparatus (Nanodrop Technology, Wilmington, DE).
Quantification of ammonia-oxidizing bacteria from the epiphytic, benthic, and pelagic samples was performed by quantitative PCR (qPCR) with the TaqMan assay, as described by Hermansson and Lindgren (20). Modifications to the concentrations of the components of the mixture in a final volume of 25 μl were as follows: half of the volume of IQ Super mix (Bio-Rad Laboratories Inc.), 1 μM each primer (CTOf and RT1), 0.5 μM probe TMP1, 200 ng/μl bovine serum albumin, and 10 ng DNA template. The reactions were performed and monitored with a thermocycler DNA engine (Opticon 2; MJ Research). Standard curves were calculated in every run on the basis of 10 known concentrations of DNA from Nitrosomonas europaea, also used in the spiking experiments to optimize the DNA concentrations in the sediment samples. To estimate cell numbers, it was assumed that β-AOB contains one rrn operon per genome. For a direct comparison between the three different compartments, the numbers of cells were expressed as the numbers of 16S rRNA gene copies/ml in the first 5 cm of the sediment by considering the densities of the sediment samples for the benthic compartment and the amount of plant biomass per volume of lake water at every station for the epiphytic compartment. Hence, in the latter case, the contribution of epiphytic β-AOB was expressed on the basis of the total amount of plant biomass in the water layer.
The compositions of the ammonia-oxidizing bacterial communities in the benthic, pelagic, and epiphytic compartments were assessed by a nested PCR-denaturing gradient gel electrophoresis (PCR-DGGE) approach involving two primer sets specific for the amplification of DNA from ammonia-oxidizing Betaproteobacteria. The method has been described in detail by Coci et al. (12).
The DGGE gel images were analyzed with Phoretix 1D advanced software (version 5.1; Non Lynear Dynamic Ltd., Newcastle upon Tyne, United Kingdom). Finally, three matrices were calculated separately for the epiphytic, benthic, and pelagic compartments, with each matrix including triplicate samples per station. Analyses were conducted with the whole set of samples, including the ones in which no β-AOB-related DGGE bands were detected. The percent distributions of the bands were calculated with Phoretix software under the assumption that a value of 100% represents the total number of bands recovered per compartment. The matrices were subsequently subjected to hierarchical cluster and statistical analyses with PRIMER software (version 5) (11). The effects of the factors compartment, lake, station, and exposure to southwestern wind on the compositions of the β-AOB communities were tested by an analysis of similarities (ANOSIM) procedure (9). The ANOSIM procedure is based on a triangular similarity matrix, in this case, of the percentage distributions of bands in the different compartments and lakes. The ANOSIM statistic R, which equals (B − W)/[n(n − 1)/4], calculates the differences between the average of rank dissimilarities arising from all pairs of replicates of the band percentage profiles between different compartments or lakes (B) and the average of all rank dissimilarities of the band percentage profiles within each compartment or lake (W). R is scaled within the range of −1 to +1. An R value of 1 means that all replicates within one compartment or lake are more similar to each other than any replicate from different compartments or lakes; an R value of 0 indicates that similarities between and within compartments or lakes are the same, on average; an R value of <0 means that the similarities across different compartments or lakes are higher than those within compartments or lakes, which usually is likely due to the incorrect labeling of samples (11). Global R gives the R value for the whole set of data collected for the compartments or the lakes. If the global R value is significant, pairwise procedures were applied to test the main between-group differences.
Nonmetric multidimensional scaling analysis was performed for different clusters and lineages of ammonia-oxidizing bacteria, including the samples in which only β-AOB-related bands were detected. To relate the β-AOB community compositions of different compartments to their respective environmental variables, we used the BIO-ENV procedure of the PRIMER software package. Briefly, the procedure calculates the Spearman's rank correlation coefficient (ρ) between the distances of the response matrix (in our case, the Bray-Curtis similarity matrices of the β-AOB communities of the three compartments) with normalized Euclidean distance matrices of appropriately transformed environmental parameters associated with the compartments (10). We used the BIO-ENV procedure to calculate the ρ value of every possible combination of predictor variables and the best fit, i.e., the single variable or the combination of variables that simply best explained the composition of the ammonia-oxidizing bacterial communities. We present only the highest ρ values and their associated combinations of environmental variables. The sequences obtained from the DGGE bands were aligned with published 16S rRNA gene sequences of cultured ammonia-oxidizing bacteria by use of the Fast Aligner tool of the ARB software, and a neighbor-joining tree was created by using the parsimony criterion and filters created ad hoc (36).
The fluorescent in situ hybridization (FISH) technique was used to detect β-AOB on the leaves of submerged macrophytes. Samples were fixed in 4% paraformaldehyde solution. Small pieces of macrophyte leaves were placed on Teflon-coated slides and the slides were embedded in approximately 10 μl of 0.1% agarose solution. The following oligonucleotide probes were used in combination: probe NSV443, which is specific for the Nitrosovibrio-Nitrosolobus-Nitrosospira group; probes NSO190 and NSO1225, which cover all sequences of ammonia oxidizers of the beta subclass (41); and probe BET42a, which is specific for all Betaproteobacteria (37). Oligonucleotides were synthesized and fluorescently labeled at the 5′ end with a hydrophilic sulfoindocyanine dye, Cy3 or Cy5, or with 5(6)-carboxyfluorescein-N-hydroxysuccinimide ester (FLUOS; Thermohybaid Inc., Germany). The hybridization steps were performed as described by Manz et al. (37). When probe NSO190 (which requires 55% [vol/vol] formamide in the hybridization buffer) was used simultaneously with any of the other probes (probes NSV443, NSO1225, and BET42a, which require 35% [vol/vol] formamide in the hybridization buffer), a successive hybridization protocol was applied (60). Detailed information on the probes used in this study can be found in probeBase (35). The slides were briefly washed with double-distilled H2O, air dried, and embedded with Vectashield mounting medium containing 4′,6- diamidino-2-phenylindole stain. For the screening, acquisition, and analysis of the images, an Axioplan 2 epifluorescence microscope (Zeiss, Jena, Germany) together with the Axiovision software package (release 3.2; Zeiss) and a DMIRE2 confocal laser scanning microscope (Leica Mycrosystem, Germany) together with Confocal software (version 2.61; Leica Mycrosystem) were used. All image analyses were performed with the standard microscope software packages.
Statistical analyses of the PAA and qPCR assays were performed with the software package STATISTICA (version 7; Statsoft., Inc.). The data were checked for normality by means of the Shapiro-Wilk test and by using Levene's test for the homogeneity of variances. In case of normality and the homogeneity of variances, a multifactorial analysis of variance (ANOVA) was used to test for the effects of the factors lake and compartment on potential ammonia-oxidizing activities and numbers of genes. Tukey's honestly significant difference tests were performed to calculate significant differences between the mean values. In the case of the absence of normality or the homogeneity of variances, a nonparametric Kruskal-Wallis ANOVA was applied to look for the effects of lakes and compartments. Correlations between the measured variables and potential activity measurements or numbers of β-AOB in sediment samples were determined by means of Spearman rank correlation tests.
The partial 16S rRNA gene sequences obtained from the DGGE bands were submitted to the EMBL data bank under the following accession numbers: AM503957 to AM503962.
Whereas the determination of PAA was possible with incubations of sediment and macrophyte samples, no NO2− plus NO3− production from ammonium could be observed during incubation of the filters containing suspended materials from the water layer (Fig. (Fig.1).1). The average values for PAA with incubations of macrophyte leaves of P. pectinatus collected from Lakes Vossemeer and Gooimeer were 4.9 ± 0.7 (standard error [SE]) and 7.8 ± 0.4 mmol NO2− plus NO3− m−2 h−1, respectively. Due to the large variation in activities at the different stations, especially in Lake Vossemeer, the average values were not significantly different between the two lakes. PAA was not detectable by the incubation of leaves of C. aspera collected from Lake Nuldernauw.
The average values of the benthic potential ammonia-oxidizing activities ranged from 7.6 ± 0.2 (SE) in Lake Nuldernauw to 16.9 ± 2.8 mmol NO2− plus NO3− m−2 h−1 in Lake Vossemeer (Fig. (Fig.11).
The lakes had a significant effect on the PAAs averaged over the three compartments (one-way ANOVA, F = 13.852, P = 0.00023). Significantly (P < 0.05) lower values were observed in Lake Nuldernauw than in the other lakes. According to a nonparametric Kruskal-Wallis ANOVA, the compartments also significantly (H = 20.10531, P < 0.0001) affected the PAAs when the PAAs were averaged over the three lakes, with the lowest values being obtained for the pelagic compartment. The benthic and epiphytic PAAs were significantly (R2 = 0.989, P < 0.005) and positively correlated with each other. The benthic and epiphytic PAAs were also significantly (P < 0.05) and positively correlated with the water retention time (R2 = 0.708 and 0.696, respectively), the Kjeldahl N content (R2 = 0.775 and 0.766, respectively), and the total P content (R2 = 0.689 and 0.678, respectively) in the water layer. The water retention time itself was also significantly (P < 0.05) and positively correlated with the Kjeldahl N content (R2 = 0.787) and the total P content (R2 = 0.997) in the water layer. No other significant correlations with physical or chemical parameters were observed for the benthic and epiphytic PAAs.
16S rRNA genes related to ammonia-oxidizing Betaproteobacteria were detected in all compartments (Fig. (Fig.2).2). On the basis of the numbers of m2 of water surface, slightly lower average numbers were observed in the pelagic compartment. Nevertheless, on the basis of the nonparametric Kruskal-Wallis ANOVA, the log numbers of gene copies averaged over the three lakes were not significantly affected by the compartment. However, according to the same statistical analysis, the logs of the gene copy numbers were significantly (H = 7.786327, P = 0.0204) larger in Lake Gooimeer than in Lake Nuldernauw when the numbers were averaged over the three compartments. No significant difference in gene copy number was observed between Lake Vossemeer and the two other lakes. The logs of the gene copy numbers in the three compartments were significantly (P < 0.05) and positively mutually correlated. No significant correlations were observed between the log of the gene copy numbers and any other physical or chemical parameter.
By means of 16S rRNA PCR-DGGE analyses, a total of two bands, three bands, and one band that belonged to the β-AOB were recovered in the epiphytic, benthic, and pelagic compartments, respectively. These bands were included in the community analysis. The affiliation of the bands with the β-AOB species is shown in a phylogenetic tree (Fig. (Fig.3).3). Members of the Nitrosomonas oligotropha lineage were detected in all three compartments. Members of cluster 3 of the Nitrosospira lineage were detected in all of the epiphytic and benthic compartments. Members of cluster 0 of the Nitrosospira lineage were detected only in the benthic compartments of Lake Gooimeer and Lake Vossemeer. No ammonia oxidizers were detected by 16S rRNA PCR-DGGE in the pelagic or epiphytic samples of Lake Nuldernauw or in the epiphytic samples of station C of Lake Vossemeer.
The triplicate samples per lake showed generally consistent band patterns in the three compartments, with the coefficients of variance calculated for the proportions of maximum bands ranging from 0% to a maximum of 10%. The results of the cluster analyses (Fig. (Fig.4)4) demonstrated that samples from the benthic compartments (indicated by the suffix B in Fig. Fig.4)4) were more than 70% similar in their compositions of β-AOB communities, while samples from the pelagic compartments (suffix P) of Lakes Gooimeer and Vossemeer were more than 90% similar. The composition of the ammonia-oxidizing community of epiphytic samples (suffix E) differed more strongly between lakes. The epiphytic samples of Lake Gooimeer were 95% similar to the benthic samples of the same lake, while the epiphytic samples of Lake Vossemeer grouped together remotely from the other samples.
To compare the community compositions between compartments, nonmetric multidimensional scaling analyses were applied. Samples in which no ammonia-oxidizing bacteria were retrieved by DGGE were also taken into account in the analysis. Analyses of similarity (Table (Table4)4) revealed an overall significant difference between compartments with respect to the ammonia-oxidizing communities (global R = 0.348, P = 0.001). The R values for pairwise tests (Rpw) indicated a higher level of significant dissimilarity between the benthic and the pelagic communities of β-AOB than between the benthic and the epiphytic compartments. The epiphytic community was not significantly different from the pelagic one. However, by considering the P value at the edge of significance (P = 0.063) obtained by the latter pairwise comparison, we repeated the analyses by excluding the pelagic and epiphytic samples in which no β-AOB-related bands were detected. In that case, the pelagic and epiphytic β-AOB communities differed significantly from each other (Rpw = 0.62, P = 0.002; data not shown in Table Table4).4). An overall significant dissimilarity between the ammonia-oxidizing bacterial communities of the different lakes was also obtained (Table (Table4).4). In particular, the communities of Lake Nuldernauw were more dissimilar from the communities of Lake Gooimeer (Rpw = 0.46) than from the communities of Lake Vossemeer (Rpw = 0.26). The latter two communities turned out not to be significantly dissimilar.
Among the physical variables, the composition of the benthic ammonia-oxidizing bacterial community was mostly correlated with the percentage of clay (ρ = 0.31) and the phosphate concentration in the sediment (ρ = 0.51). The pelagic and epiphytic β-AOB community compositions were both mostly correlated with the concentration of total nitrogen (ρ = 0.39 and 0.59, respectively) in the water layer.
To confirm the presence of ammonia-oxidizing Betaproteobacteria on the leaves of submerged plants, fluorescent in situ hybridization was applied to the leaf samples. Detection of ammonia-oxidizing bacterial cells on the leaves of P. pectinatus was difficult by means of epifluorescence microscopy due to the autofluorescence of the macrophyte surfaces. By means of confocal laser microscopy, cells belonging to the Nitrosospira lineage were observed on the leaves of P. pectinatus at stations A, B, and C of Lakes Gooimeer and Vossemeer (data not shown). Cells belonging to the N. oligotropha lineage were never observed on the leaves of P. pectinatus from these stations. No ammonia-oxidizing bacterium-like cells were observed on C. aspera, which dominated the stations of Lake Nuldernauw.
The epiphyton of submerged macrophytes represents a niche for bacterial colonization in shallow freshwater lakes (61). In a recent, qualitative study of the bacterial ammonia-oxidizing community composition that had been conducted in the same system of interconnected shallow freshwater lakes used for the present study, we demonstrated the presence of β-AOB on the leaves of submersed macrophytes by means of the PCR-DGGE technique based on the 16S rRNA gene. We showed that the epiphytic, benthic, and pelagic communities significantly differed from each other (12). Those data were confirmed by the findings of the present study and were completed with quantitative data on the numbers of 16S rRNA gene copies of β-AOB cells in the three compartments obtained by qPCR and with direct microscopic images.
The numbers of ammonia-oxidizing bacterial cells were not significantly different between the compartments. The numbers of ammonia-oxidizing bacteria are usually smaller in the pelagic compartment than in the benthic compartment, when they are expressed per ml of water (29, 52, 62). However, integration of the numbers for the whole water column increased their population to a size comparable to that of the population present in the upper 5 cm of the sediment. The numbers of pelagic cells were similar in the different lakes, with a higher degree of variability being seen in Lake Vossemeer. The shallowness might have been responsible for the resuspension of sediment particles and, consequently, of particle-attached bacteria.
The potential ammonia-oxidizing activity per cell could be calculated by assuming that there was one 16S rRNA gene copy per cell (Fig. (Fig.5).5). The values ranged from 0 in the pelagic compartment to 111 nmol cell−1 h−1 in the epiphytic compartment. However, the variation in activity per cell was high, as can be seen from the standard deviation bars in Fig. Fig.5.5. According to the Kruskal-Wallis ANOVA, the potential activities per cell were not significant different for the lakes, but the activities per cell in the pelagic compartment were significantly (P < 0.05) different from the activities per cell in the benthic and epiphytic compartments (H = 16.980856, P = 0.0002). The large variation in potential activity per cell between the compartments is most likely determined by the environmental conditions under which the cells reside. Being attached either to sediment particles or submerged macrophytes seems to be favorable for the ammonia-oxidizing cells.
Of particular interest is the difference in the numbers of β-AOB cells colonizing the two different submerged plants investigated, namely, P. pectinatus and C. aspera. P. pectinatus is a vascular plant with a high shoot surface area, easily accessible by microorganisms for attachment. Individuals of C. aspera are anatomically highly developed green algae with a lower surface area-to-volume ratio. Chara species contain high concentrations of sulfur compounds (1), which likely inhibit nitrification (25). At present, only a few studies of the β-AOB associated with different plant species are known (4-6). However, none of them are related to the epiphyton in freshwater environments, which prevents a direct comparison with our results. However, the consequences of a specific association of bacteria with particular macrophytes might be suggested by our potential activity measurements that again showed the nonsupporting role of charophytes for β-AOB. Eriksson and Andersson (16) reported on a difference in nitrifying potential activities between macrophyte species and suggested that differences in the physical and chemical characteristics of the plants themselves, as well as the effects of the plants on environmental variables within macrophyte stands, caused the dissimilarities observed. Our results are supportive of this suggestion, since the presence of particular species of macrophytes is both a cause and a consequence of the trophic status, as well as the hydrodynamics of the lakes. Charophytes were introduced into Lake Nuldernauw during the restoration process because of their strong positive effect on water transparency (56). Studies on nitrogen uptake by charophytes (58) indicated their preferential uptake of NH4+ rather than NO3− from both the below- and aboveground parts of the macroalga, evoking a possible competition for NH4+ with epiphytic β-AOB. Moreover, a significant depletion of NH4+ from the sediment caused by Chara spp. (58) also suggested competition for nitrogen with benthic ammonia oxidizers. This might contribute to the significantly lower PAA rate measured in the benthic samples of Lake Nuldernauw. In our study, potential ammonia-oxidizing activities were significantly correlated with the nutrient status of the water layer, as indicated by the amounts of Kjeldahl nitrogen and total phosphorus.
The epiphytic β-AOB communities differed not only between macrophytes species but also between lakes. Members of cluster 3 of the Nitrosospira lineage appeared to be the most common inhabitants of the epiphyton of P. pectinatus in most of the lake stations sampled. Members of this cluster are predominantly found in soils but also occur in freshwater environments and wastewater treatment plants (cf. reference 27). For example, they have also been found to be among the dominant β-AOB in the littoral zone of Lake Drontermeer, which is located between Lake Vossemeer and Lake Nuldernauw (30). Members of the N. oligotropha lineage were detected only in the epiphytic samples of Lake Gooimeer. The dominance of members of the N. oligotropha lineage in the pelagic compartment is in line with the isolation of various members of this lineage from freshwater habitats (28, 52), as well as with the findings of other studies conducted in the same sampling area (51).
Finally, members of cluster 0 of the Nitrosospira lineage were exclusively detected in the benthic compartments of both Lake Gooimeer and Lake Vossemeer. Ammonia oxidizers of cluster 0 of the Nitrosospira lineage (48) were mainly detected in undisturbed and unfertilized soils (7, 33), but they have also been retrieved from the same sampling area by Speksnijder et al. (51).
The analysis of the composition of the ammonia-oxidizing Betaproteobacteria community was done at the time that the standing biomass of the macrophytes in the lakes was at the maximum. Hence, the analyses present measurements for only a single time point, and the distribution of microbial species could have been different in other seasons. However, in earlier studies concerning the distribution of ammonia-oxidizing Betaproteobacteria in the water column and in the sediment of an intertidal freshwater marsh in the Scheldt estuary, we were not able to detect a significant effect of the season on the distribution of species (14, 33a).
Both measurement of PAAs and determination of 16S rRNA gene copy numbers are meant to estimate the number of ammonia-oxidizing cells. The fact that we were not able to find a significant correlation between the results of these two methods might be explained in two ways. First, PAA is meant to estimate the number of actively cells oxidizing ammonia, whereas detection of the 16S rRNA gene points to the presence of active and nonactive cells. Second, PAA may involve the participation of ammonia-oxidizing Crenarchaea, whereas the primer set used to enumerate the 16S rRNA genes is directed only to ammonia-oxidizing Betaproteobacteria. Recently, ammonia-oxidizing Crenarchaea were found to dominate the ammonia-oxidizing communities in the rhizosphere of freshwater macrophytes (21), indicating a possible role of the archaeal component in freshwater ecosystems.
Although no potential ammonia-oxidizing bacteria could be measured in the samples from the pelagic compartment, β-AOB were apparently present, as shown by the use of real-time PCR with a 16S rRNA gene-specific primer set. No β-AOB were detected by means of PCR-DGGE analyses of the samples from the pelagic and epiphytic compartments of Lake Nuldernauw (Fig. (Fig.4),4), and no potential activities were measured in these samples as well (Table (Table4),4), but the qPCR method applied detected β-AOB in both compartments of this lake. This shows the superiority of the qPCR method for the detection of low numbers of ammonia-oxidizing cells. Whereas qPCR and 16S rRNA-based PCR-DGGE aim at identifying the total numbers of ammonia-oxidizing cells, i.e., active and nonactive cells, potential activity measurements zoom in on active cells. The present study also provides the first report of the presence of β-AOB on the leaves of submerged P. pectinatus macrophytes by microscopic analysis. However, whereas the PCR-DGGE method indicated the presence of members of both the N. oligotropha lineage and cluster 3 of the Nitrosospira lineage, only cells of the latter lineage were observed by the FISH method. In addition, when the qPCR method suggested the presence of β-AOB on the leaves of C. aspera, no cells could be detected by the FISH technique in the epiphyton of this macroalga. This could have been due to the limitations of the FISH method used. Higher-quality images could have been obtained by applying the catalyzed reporter deposition-FISH protocol (55).
The overall condition of the lakes appeared to be of importance for the diversity of the communities of ammonia-oxidizing Betaproteobacteria. The benthic community structure of β-AOB was mostly correlated with the clay content and the phosphate concentration in the sediment, whereas the pelagic and epiphytic community structures were mostly correlated with the total nitrogen concentration in the water layer. The communities of the most restored lake, i.e., Lake Nuldernauw, were poorer in terms of species richness than the communities of Lakes Gooimeer and Vossemeer. The sediment composition, for example, had a direct effect on the diversity of β-AOB, which was already known from the development of wetlands constructed for the removal of nitrogen from wastewater (19). The adsorption of ammonia by clay minerals may provide localized surface-associated ammonia oxidation (46), thereby enhancing the nitrification process. In our case, Lake Nuldernauw, which has the more sandy sediment, showed a lower level of diversity.
Ammonia-oxidizing Betaproteobacteria cells clearly inhabit the epiphyton of submerged macrophytes, but their presence is dependent on the macrophyte species. Sequences of ammonia-oxidizing Betaproteobacteria present in both the benthic and the pelagic compartments were also retrieved from the leaves of P. pectinatus. The epiphytic ammonia-oxidizing bacterial community differed qualitatively and quantitatively from the benthic and pelagic communities. The role of the epiphytic ammonia-oxidizing bacterial communities in the internal nitrogen turnover in shallow freshwater lakes during the period when the macrophytes are present must be taken into account.
Published ahead of print on 22 January 2010.