OUP user menu

Characterisation of the culturable heterotrophic bacterial community in a small eutrophic lake (Priest Pot)

Mary L. Edwards, Andrew K. Lilley, Tracey H. Timms-Wilson, Ian P. Thompson, Ian Cooper
DOI: http://dx.doi.org/10.1111/j.1574-6941.2001.tb00815.x 295-304 First published online: 1 May 2001


The community composition and structure of planktonic heterotrophic bacteria (903 isolates) sampled from a small eutrophic lake in northern England (Priest Pot) was studied with respect to season (four samples) and depth (to 3.1 m). Bacteria (887) were isolated on tryptic soy broth agar and identified to 48 genera using fatty acid methyl ester analysis. The two most abundant genera isolated were Aeromonas and Pseudomonas which, respectively, dominated the middle to bottom depths in August and all depths in February. The structure of the sampled community was described using: species richness, Simpson's index and the Shannon–Wiener index. All three indices detected a number of significant differences with depth demonstrating stratification. The greatest stratification of the bacterial community was observed in August when bacterial counts correlated strongly and negatively with diversity. Using structural measures was found to be preferable to the use of species frequencies in the analysis of perturbation and succession in community structure. Insensitivity to one or more of eight antibiotics was observed in 71% (61/86) of the isolates tested particularly in Gram-negative genera. Bacteriocinogeny and lysogeny was observed in 36% (32/90) of isolates. Using sensitive indicator strains, two of 10 producing strains produced virus, while the others produced bacteriocins.

  • Fatty acid methyl ester microbial identification system
  • Bacterial diversity
  • Freshwater lake
  • Bacteriocin
  • Bacteriophage
  • Antibiotic resistance

1 Introduction

The study of bacterial diversity raises substantial challenges of rapid and reliable characterisation, and representative sampling. A wholly reliable inventory of diversity present is still not completely within reach. To improve understanding of diversity we used the Microbial Identification System (MIS), to identify large numbers (903) of bacterial isolates by comparing their methylated fatty acid profiles, with reference to a substantial database (>70 000 isolates). MIS is a semi-automated commercially available system, used routinely to identify bacteria in many hospitals, and widely applied to examine the culturable bacteria isolated from soil or plant samples [1,2].

Priest Pot at the north end of Esthwaite Water in the English Lake District, Cumbria is a small lake that is isothermal and aerobic in the winter. In the summer it becomes thermally stratified with an oxic epilimnion, a microaerobic metalimnion and an anoxic hypolimnion. Previous studies investigated the diversity of microbe communities in Priest Pot [3], but concentrated on protozoa [4] and rotifers [5]. Virus-like particles have been observed in the water [3] and interactions between ciliates, bacteria and virus-like particles have been inferred, using electron microscopy [6]: virus-like particles have not been linked to specific bacteria. This study therefore adds to the evolving holistic picture [7] of the ecology of Priest Pot.

Our principal aim was to investigate the relationship between biotic and abiotic factors and bacterial diversity and succession. For this, it was important to have reliable indicators of change in the sampled bacterial community. We have applied structural indices which proved suitable for statistical analysis with the samples collected. We have considered antibiotic resistances because they reveal groups within a community, and interactions with natural antibiotic producers, potentially maintaining community structure in natural habitats and acting as reservoirs for resistance genes capable of transfer into vertebrate (human and veterinary) pathogens.

2 Materials and methods

2.1 The study site and physico-chemical measurements

The study site was a small (1 ha) lake with a maximum depth of 3.6 m located at 54°22′ N, 2°59′ W and surrounded by willow and alder carr. Priest Pot is fed by groundwater, field drains and two or three slow-flowing ditches, from improved grassland and rough grazing. Oxygen and temperature were determined in situ, with a stirrer-assisted Yellow Springs instrument.

2.2 Bacteria

Water was taken from the pond using a pneumatically operated sampler [8]. Samples (10 ml) were taken on 4 June 1997, 30 June 1997, 20 August 1997 and 12 February 1998, from each of 12 depths. Within 24 h, water (0.1 ml) was spread on tryptic soy broth agar (TSBA: Oxoid) containing cycloheximide (50 mg l−1) and the plates were incubated at 25°C for 48 h. Colonies were selected randomly. Three hundred isolates were collected on 30 June, 20 August and 12 February and 100 on 4 June. Purified isolates were stored with saline glycerol at −70°C. It is recognised that TSBA introduces a bias toward fast-growing aerobic heterotrophs, however, this medium facilitates comparison of our results with other studies we have made on the same medium. We are currently comparing the structure and diversity of bacterial communities in the rhizosphere, phyloplane, an industrial bioreactor, 1,2-dichlorobenzene-polluted soil, Priest Pot and groundwater aquifers. The use of TSBA enables comparisons between samples treated in the same way.

Bacteria were identified by fatty acid methyl ester (FAME) analysis. FAME extracts were prepared and analysed by gas chromatography using the MIDI-MIS, as described previously [9]. Using a pattern recognition algorithm, FAME profiles were matched to those in the MIDI-MIS TSBA aerobic data base (ver. 3.3). The MIS data base's closest matches were tested by cluster analysis to confirm the separation of isolates into distinct species, using a 6-Euclid cut-off point [2,9].

2.3 Measurement and assessment of bacterial community structure

From our experience of indices of change in population structure of bacterial communities, three methods that proved suitable were applied.

  1. Species richness (S*). This is simply the number of species in a sample.

  2. Simpson's index (L′). Formally L′ is the probability that two isolates taken at random from a sample will be the same species. Of the three measures, this is the most sensitive to changes in the frequency of the more abundant species.

  3. Shannon–Wiener index (H′). This is a combined figure that summarises the extent of diversity and the evenness of isolate distribution within taxa. It is sensitive to changes in the frequency of common and less common but not rarer species.

These indices were calculated by: Embedded Image Embedded Image (ni=number of species i in the sample, s=number of species in the sample, pi=proportion of species i in the sample, N=number of individuals in the sample).

Differences in S*, L′ and H′ were tested using the method of Solow [10]. This employs a randomisation test as follows: isolates for two samples (A and B) are listed, the chosen population parameter (for example, H′) calculated and the difference in the parameter for the two samples noted (DA−B=H′A−H′B). The isolates in the two samples are then amalgamated, randomly mixed and repartitioned to the original sample sizes, the population parameter re-calculated and a new difference noted. This random mixing, repartitioning and calculation of the difference is repeated 10 000 times and the values (D1 to D10 000) ranked. Finally, to test if H′A and H′B are significantly different, at the 95% confidence level, DA−B is compared with the ranked values. If this difference is greater than or equal to the value of the difference at the 97.5 percentile, or less than or equal to the value of the difference at the 2.5 percentile, then the values for the parameter of the two samples are considered significantly different. An Excel macro-program was written to apply Solow's method to the pairwise comparison of each of the three parameters (S*, L′ and H′), using 10 000 re-samples per test.

2.4 Antibiotic sensitivity

To test for antibiotic sensitivity, bacteria cultured on TSBA were transferred to tryptic soy broth (TSB) and grown to an absorbance of 0.4–0.6 at λ 600 nm, mixed with soft agar (0.7% w/v) and overlaid onto TSBA. Antibiotics (ampicillin, 10 μg; cefepine, 30 μg; norfloxacin, 10 μg; erythromycin, 15 μg; gentamicin, 10 μg; neomycin, 30 μg; penicillin G, 10 μg; and tetracycline, 30 μg), contained in 6-mm filter paper discs, were dropped onto the plates, using a susceptibility testing kit (Oxoid HPO53A). The plates were incubated overnight at 28°C, then observed for growth inhibition.

2.5 Particle production

To test for bacteriocin or phage production, bacteria were grown to an absorbance of 0.4–0.6. at λ 600 nm. Mitomycin C (Sigma) was added to give a final concentration of 0.1–0.2 μg ml−1 and the cultures were incubated overnight at 28°C along with equivalent untreated cultures. Cultures with absorbance values less than their untreated counterparts were centrifuged at low speed to remove cells and cell debris. The supernatants were negatively stained with 1% aqueous uranyl acetate and examined in a transmission electron microscope at a magnification of 300 000×.

2.6 Sensitive isolates and virus/bacteriocin differentiation

To test for host sensitivity, bacteria were grown in TSB to an absorbance of 0.4–0.6 at λ 600 nm. Cells (0.1 ml) were mixed with 2.5 ml of soft TSBA (agar 0.75% w/v) and layered onto TSBA plates. Filtrates (0.22 μm Corning filters) of mitomycin C-induced cultures were applied to the plates, using a glass rod replicator. Plates were incubated overnight and assessed for inhibition of growth. To differentiate virus from bacteriocin, transfers were made from areas of inhibition into 2 ml of medium. The medium was filtered to remove cells and applied (10 μl) to freshly prepared plates of the original host bacterium. When growth inhibition was maintained, after two transfers, the inhibitor was judged to be a virus rather than a bacteriocin.

3 Results

3.1 Description of the water column

To relate the bacterial community to other water column features, measurements of temperature and oxygen were made at different depths. In the summer, the water column was stratified with a warm aerobic zone in the upper region of the column and a colder anoxic zone in the lower. On 4 June, the maximum temperature was 18°C at the top of the column, declining to 9°C at the bottom. Epilimnetic oxygen concentration was 12–17 g l−1 and the oxic–anoxic boundary was at 2–2.25 m.

Conditions on 30 June differed from those on 4 June and this was probably due, in part, to winds and low air temperatures in the previous week. The maximum temperature was only 15°C and the difference in temperature between the top and bottom of the column was 5°C, compared to 9°C on the earlier sampling date. The position of the oxic–anoxic boundary was approximately the same as on 4 June.

The highest temperatures and shallowest oxic–anoxic boundary were recorded on the August sampling date; the maximum temperature was 22°C, at the top of the column, and the minimum, 11°C, at the bottom; the aerobic zone extended to only 1.5 m. Conditions in the water column were not measured on the 12 February sampling date. However, from observations in 1997 we know that the water column at this time of the year was isothermic (5°C) and entirely aerobic (9–11 mg l−1).

3.2 Distribution of bacteria as measured by colony-forming units

Counts of colony-forming units (cfu) varied with sampling date and depth. On 4 June, 30 June and 12 February, counts were low, ranging from 40 to 770, 50 to 890 and 110 to 1390 cfu ml−1 respectively, compared to higher counts on 20 August that ranged from 700 to 22 200 cfu ml−1. In June, counts did not vary in an obvious pattern with depth. In February, counts were greatest near the surface (750 cfu ml−1 at 0 m) and just above the sediment (1390 cfu ml−1 at 3.2 m). In August, there was a clear peak at 2 m beginning in the oxic region and extending into the anoxic zone. This peak coincided with the highest concentration of total phosphorus and with one peak of soluble reactive phosphorus (data not shown); no other coincidences were noted.

3.3 FAME identification of the bacteria

Of the 903 isolates analysed using the MIDI-MIS FAME system, 887 were identified. The identified isolates were assigned to 48 genera (32 Gram-negative and 16 Gram-positive) and 99 species (Table 1). Of the 48 genera identified two, Pseudomonas and Aeromonas, constituted 50% of the isolates; 15 genera, present at a frequency of 1% or greater, accounted for 39%; a further 22 genera accounted for 9.3% and 10 genera, each represented by a single isolate, accounted for 1.7%.

View this table:

Abundances of bacterial genera sampled from Priest Pot on four occasions and at various depths

Sample nameGenera%Sample nameGenera%Sample nameGenera%Sample nameGenera%
N=34Aeromonas18 N=47Pseudomonas15 N=40Chryseobacterium18 N=93Acinetobacter26
Vogesella6Aeromonas4Neisseria5No match2
Jun04-MAeromonas32Bacillus6Salmonella3 N=36Bacillus11
Brevundimonas5Aeromonas2 N=46Chryseobacterium7Micrococcus3
No match3Enterococcus2Feb12-MMPseudomonas73
Stenotrophomonas3Hydrogenophaga2Aug20-MMAeromonas100 N=26Bacillus8
Yersinia3Serratia2 N=50Janthinobacterium4
Jun04-BPseudomonas26Jun30-MBPseudomonas55 N=49Acinetobacter14Salmonella4
N=42Aeromonas19 N=55Brevundimonas11Serratia8Serratia4
Micrococcus5Aeromonas4Pseudomonas2 N=58Bacillus12
No match5No match4Salmonella2Hydrogenophaga7
Nocardia2Microbacterium2 N=66Bacillus8Flavobacterium3
Aeromonas14No match3Paenibacillus2
No match3Brevibacillus2 N=89Acinetobacter17
Listonella2Enterobacter2No match4
  • N is the number of isolates. T, top; MU, middle upper; MM, middle middle; MB, middle bottom; M, middle; B, bottom. Sampling dates were 4 June, 30 June, 20 August and 12 February. See also Table 2.

3.4 Comparison of sample composition

To facilitate comparisons, data from adjacent depth aliquots were amalgamated to produce a small number (three to five) of depth range samples, avoiding the pooling of samples with very different species patterns, yet providing sample sizes large enough for analysis (see Table 2 for details). In Table 1, all genera identified are listed and ranked according to their abundance in each depth range sample.

View this table:

Diversity measure (S*, L′ and H′) values estimated for Priest Pot samples and adjusted for uniform sample sizes of 30 isolates

Depth range (m)Sample nameaEstimated values (S*, L′ and H′) when sample=30 isolates
  • Each value was estimated with a re-sample size of 30 isolates and taking the median from 1000 re-samplings.

  • a a Sample name indicates position in water column and date sampled: T, top; MU, middle upper; MM, middle middle; MB, middle bottom; M, middle; B, bottom. Sampling dates indicated are 4 June, 30 June, 20 August and 12 February.

Not surprisingly, when ranked positions were considered, all 17 samples had different compositions (Table 1). However, differences in sample composition were also apparent when only the most dominant genus in each sample was examined; in all instances this was one of six genera (Pseudomonas, Aeromonas, Brevundimonas, Exiguobacterium, Acinetobacter and Bacillus). Inclusion of the second most dominant group resulted in the addition of one genus, and the third, a further eight. Similarity in composition, with respect to the three most dominant genera, occurred in only two instances (20 August, top and middle upper; 12 February, top and bottom).

Only Aeromonas (20 August) showed an obvious distribution pattern with depth. Numbers of Aeromonas isolates were small at the top of the column, representing 15% of the colonies sampled, increased to a peak at 1.9 m where they represented 100% and declined at the bottom to 56% of the isolates. In contrast, Pseudomonas, the other most commonly isolated genus, was distributed evenly down the column on 4 June.

3.5 Comparison of sample structures

To examine community diversity in more detail, we compared structures, using the indices of S*, L′ and H′, calculated for samples taken at different depths. Values for these parameters were estimated for a notional sample size of n=30(Table 2) using a re-sampling method with 1000 re-samples per estimation. Inspection of these values (Table 2) revealed that the most diverse samples were from the bottom of the water column, with the August sample being the least diverse.

To test whether differences in diversity were significant, depth ranges were compared with respect to S*, L′ and H′, using Solow's test (Table 3). This tests whether the differences in S*, L′ or H′ for two samples are the natural random product of sampling (allowing for differences in sample size) or are the result of real differences in the distribution of bacteria, isolated from the depths sampled.

View this table:

Matrices of significance values from pairwise tests of differences in diversity between samples from the water column on three dates

Species richness (S*)
Feb12-T0.5670.134 0.004 0.004
Feb12-MU0.412 0.026 0.020
Jun30-T0.2900.152 0.028
Jun30-MU0.589 0.008
Jun30-MB 0.004
Aug20-T 0.002 0.002 0.006 0.561
Aug20-MU 0.002 0.046 0.002
Aug20-MM 0.002 0.002
Aug20-MB 0.006
Simpson's index (L′)
Feb12-T0.2320.705 0.004 0.002
Feb12-MU0.2000.404 0.022
Feb12-MM 0.030 0.002
Jun30-T0.172 0.004 0.002
Jun30-MU0.206 0.002
Jun30-MB 0.002
Aug20-T 0.002 0.002 0.002 0.446
Aug20-MU 0.002 0.883 0.004
Aug20-MM 0.008 0.002
Aug20-MB 0.032
Shannon–Wiener index (H′)
Feb12-T0.3980.653 0.004 0.002
Feb12-MU0.7730.098 0.018
Feb12-MM0.090 0.014
Jun30-T0.1340.066 0.002
Jun30-MU0.779 0.002
Jun30-MB 0.002
Aug20-T 0.002 0.002 0.006 0.733
Aug20-MU 0.002 0.256 0.002
Aug20-MM 0.002 0.002
Aug20-MB 0.008
  • Significant differences (P=0.05) are highlighted in bold italics. The data for 4 June are omitted as no significant differences were detected. Samples are named using the nomenclature given in Table 2.

In August, the samples were grouped into five depth ranges (Table 2). Each depth range was significantly different (P<0.05), with respect to S*, L′ and H′, from its neighbouring ranges, indicating stratification (Table 3). This was the most stratified water column observed; the results are plotted in Fig. 1. Of the 35 species detected in the top and bottom ranges, 27 were unique to either the top or bottom and none of the remaining eight species were among the more numerous.


Plot, for 20 August, of Priest Pot's lake depth against diversity adjusted for a constant sample size of 30. Three measures of diversity are presented: species richness (S*), Simpson's index (L′) and the Shannon–Wiener index H′). Diversity values were estimated by taking 10 000 re-samplings from the same data set, each of 30 isolates, and taking the medians of the calculated measures.

The 30 June samples were grouped into four water column depth ranges (Table 2). The bottom depth range sample was significantly (P<0.05) different from those above it (with respect to S*, L′ and H′), which were not distinguishable from one another (Table 3). Hence a limited stratification was observed with the community at the bottom being distinct from and more diverse than in the mixed waters above.

The 4 June samples were grouped into three depth ranges (Table 2) and showed no significant differences in S*, L′ or H′ in the water column (Table 3). The lack of community stratification was assumed to be due to mixing.

The 12 February samples were grouped into five depth ranges (Table 2) and showed a tendency to stratification, but with significant differences, and increased diversity limited to the two lowest depths (Table 3).

3.6 Correlations between sample structure and parameters measured in the lake

For each sample date we sought correlations (Pearson; r) between S*, L′ and H′ and the concentration of oxygen, and the bacterial counts. The concentration of oxygen gave poor correlations that were non-significant. Colony counts, however, correlated strongly and negatively with diversity in the August sample (for S*r=−0.835,P=0.078; for L′r=0.949, P=0.014; and for H′r=−0.897, P=0.039). In the more mixed 30 June samples, no correlations with colony counts were noted (r<0.484, P>0.50). In the 4 June sample the only significant correlation was between colony counts and L′ (r=0.997, P=0.047). When the data from the three samples, 4 June, 30 June and 20 August, were combined a moderate and significant negative correlation between colony counts and diversity was observed (for S*r=−0.764, P=0.004; for L′r=0.746, P=0.005; and for H′r=−0.758, P=0.004).

3.7 Antibiotic susceptibility

Isolates (86), chosen randomly from the collection at depths of 0.5, 2.0 and 3.25 m from three dates (30 June, 20 August and 12 February), were tested to determine their sensitivity to eight antibiotics. Seventy-one per cent of the isolates were insensitive to one or more antibiotics. Multiple resistance was common. Isolates belonging to Gram-negative genera were found to be insensitive to antibiotics more frequently than those belonging to Gram-positive genera (Table 4). Generally, patterns of antibiotic resistance were specific to a genus. For example, all Aeromonas isolates (15) were resistant to penicillin G and ampicillin; 90% (18/20) of Pseudomonas isolates were resistant to penicillin G, ampicillin and erythromycin. The patterns of resistance in isolates identified as Hydrogenophaga provided a contrast: one isolate was susceptible to all eight antibiotics tested where the other was resistant to six.

View this table:

Patterns of antibiotica resistance and susceptibility among genera isolated from Priest Pot

GenusResistance patterns and number of isolates resistant to
Brevundimonas (P. vesicularis)81
  • Distribution of antibiotic-resistant isolates (86), collected on three dates from three depths and tested against eight antibiotics.

  • a a P, penicillin G; AMP, ampicillin; E, erythromycin; NOR, norfloxacin; FEP, cefepine; N, neomycin.

3.8 Bacteriocinogeny and lysogeny

Isolates tested for their response to antibiotics were also examined. Bacteria (90) were selected randomly from three depth samples (0.5, 2.0 and 3.25 m), collected on 30 June, 20 August and 12 February. Cultures were examined for particle production following mitomycin C treatment. About one third of the isolates (32/90) produced virus-like particles, visible with an electron microscope. The proportion of particle-producing isolates belonging to Gram-positive genera (45%) was greater than that for Gram-negatives (29%). Several particle morphological types were seen with the electron microscope and most were observed in isolates of more than one genus (Table 5). Sensitive bacteria were found for the products of 13 producing isolates. The host ranges for the producing isolates varied from narrow, within genus, to broad with cross-reactions between genera (Table 6). The large number of different host ranges was partly due to the high diversity of the test isolates. Nevertheless, even within a single genus (such as Pseudomonas with seven producing isolates tested against 20 possible hosts), all host ranges were different. The products of 10 producing strains, for which sensitive hosts were found, were tested to determine if they were virus or bacteriocin. Two isolates, Delftia 489 and Pseudomonas 487, produced growth inhibition that transferred, indicating a virus rather than a bacteriocin.

View this table:

Particle types produced by Priest Pot isolates of different bacterial genera

Particle morphologyIdentification (nearest match) of isolates in which morphotype was found
Rods: straight or flexibleAcinetobacter, Aeromonas, Microbacterium, Bacillus, Citrobacter, Delftia, Paenibacillus, Pseudomonas
Head and tail particlesAcinetobacter, Arthrobacter, Microbacterium, Bacillus, Hydrogenophaga, Paenibacillus, Pseudomonas
Contractile tail assembliesAeromonas, Hydrogenophaga, Neisseria, Pseudomonas
IsometricBacillus, Brevundimonas
FilamentsDelftia, Bacillus, Paenibacillus
View this table:

Matrix of inhibitory interactionsa

Sensitive isolateIsolate no.Isolates of bacteria producing inhibitory substances
Ac 875Ba 444Ba 377Pa 481De 489Ne 679Ps 233Ps 487Ps 106Ps 919Ps 1099Ps 427Ps 1111
  • Growth inhibition of a sensitive isolate is indicated by ‘yes’. Producing isolates are: Ac, Acinetobacter; Ba, Bacillus; Pa, Paenibacillus; De, Delftia; Ne, Neisseria; Ps, Pseudomonas.

4 Discussion

Thermal layering and mixing (thermal or wind) of the water combine to create a dynamic spatial context for the development of bacterial communities in Priest Pot. Abiotic factors such as temperature, light and oxygen gradients influence bacterial communities which, in turn, adapt and modify the environment. The stratification of bacteria was therefore studied as an interaction between abiotic and biotic factors.

We did not detect any simple correlation between diversity and water column abiotic features. However, diversity negatively correlated with bacterial counts. While this is certainly a complex interaction, two interpretations seem noteworthy. Firstly, elevated counts may have resulted from growth of a limited component of the community and thus were associated with reduced diversity. Secondly, reductions in counts may have involved a greater reduction of the dominant populations and thus were associated with increased diversity.

Greater bacterial numbers were observed just below the oxic–anoxic boundary, in August. Similar stratifications were reported in Priest Pot for other microorganisms; eukaryote photoautotrophs concentrate in the oxic epilimnion and the microaerobic metalimnion, photosynthetic bacteria in the anoxic hypolimnion and protozoa species concentrate at different depths [4]; zoo-chlorella-bearing ciliates aggregate at the oxic–anoxic boundary [11] as do phycodnavirus hosts (M.L. Edwards, unpublished data).

Aeromonas is an opportunistic bacterium and has been described as an r-strategist in Lake Plußsee in Germany, another eutrophic lake where it was present in high concentration in the anoxic metalimnion [12]. We found Aeromonas dominating at the same relative position in the water column in Priest Pot.

We found, as others have [1315], that a large proportion of our environmental isolates were insensitive to antibiotics. McKeon et al. [13] reported that antibiotic resistance in bacteria isolated from groundwater was associated with one particular species within a well rather than distributed among different species and well water sources. Similarly, when comparing bacteria from different water sources in Cumbria, UK [16] and Spain [15], antibiotic sensitivity showed a greater degree of relationship with particular groups and genera than with habitat. This correlation between taxon and antibiotic sensitivity suggests a genetic stability.

Strictly comparable standard conditions have not yet been established for antibiotic testing of environmental bacteria [17], but there were some noteworthy differences and similarities between our results and those of others. For example, we did not find any tetracycline-resistant isolates. By contrast, between 9 and 12% of isolates of Pseudomonas, Aeromonas and Bacillus from a lake in Spain were resistant to tetracycline [15]. In other observations, tetracycline inhibited the growth of 6% of pseudomonads isolated from an upland river catchment [14]. Also, in Lake Windermere, which is exposed to considerable human activity and near to the site of our study, high frequencies of resistance were found to oxytetracycline. Numbers of oxytetracycline-resistant isolates were especially large in the sediment under fish cages and were thought to be due to the fish feed containing tetracycline [17].

A large proportion of the tested bacteria produced viruses or bacteriocins. Interactions mediated by such agents were observed between closely related bacteria, as well as across genera. In other studies in aquatic environments [18,19], the incidence of lysogeny was not explicitly distinguished between viruses and bacteriocins, although morphological differences were acknowledged [20]. In testing the transferability of inhibitory substances, we revealed that only a minority of particle-producing isolates harboured infectious viruses; i.e. the majority of our producers produced bacteriocins. It is difficult to envisage bacteriocins having a major effect on bacterial populations free in the water, but they might have impact on bacteria aggregated on surfaces as in biofilms and flocs.

This work and studies of phytosphere bacterial communities (A.K. Lilley, unpublished data) have found that community structural indices are more reproducible and reliable measures of community change than frequencies of individual populations. Statistically reliable estimates of the frequencies of individual populations require exceedingly large samples, but the three structural parameters of compositional change considered here were sufficiently stable, at the sample sizes we used, to readily permit an analysis of succession or perturbation. Though diversity calculations ignore the identity of the populations and focus on their numerical relationships, they can be an economical and practical way of describing, comparing and analysing samples. This approach to succession and perturbation within and among communities focuses on structural rather than compositional change.


We are grateful to the staff and CEH-Windermere for obtaining the water samples from Priest Pot. This work was supported by a grant from the NERC Centre for Ecology and Hydrology Integrating Fund.


  1. [1]
  2. [2]
  3. [3]
  4. [4]
  5. [5]
  6. [6]
  7. [7]
  8. [8]
  9. [9]
  10. [10]
  11. [11]
  12. [12]
  13. [13]
  14. [14]
  15. [15]
  16. [16]
  17. [17]
  18. [18]
  19. [19]
  20. [20]
View Abstract