Differential Growth Response of Colony-Forming α- and γ-Proteobacteria in Dilution Culture and Nutrient Addition Experiments from Lake Kinneret (Israel), the Eastern Mediterranean Sea, and the Gulf of Eilat

Abstract

Even though it is widely accepted that bacterioplankton growth in lakes and marine ecosystems is determined by the trophic status of the systems, knowledge of the relationship between nutrient concentrations and growth of particular bacterial species is almost nonexistent. To address this question, we performed a series of culture experiments with water from Lake Kinneret (Israel), the eastern Mediterranean Sea, and the Gulf of Eilat (northern Red Sea). In the initial water samples, the proportion of CFU was typically <0.002% of the 4′,6′-diamidino-2-phenylindole (DAPI) counts. During incubation until the early stationary phase, the proportion of CFU increased to 20% of the DAPI counts and to 2 to 15% of the DAPI counts in unenriched lake water and seawater dilution cultures, respectively. Sequencing of the 16S ribosomal DNA of colony-forming bacteria in these cultures consistently revealed an abundance of α-proteobacteria, but notable phylogenetic differences were found at the genus level. Marine dilution cultures were dominated by bacteria in the Roseobacter clade, while lake dilution cultures were dominated by bacteria affiliated with the genera Sphingomonas and Caulobacter. In nutrient (glucose, ammonium, phosphate) addition experiments the CFU comprised 20 to 83% of the newly grown cells. In these incubation experiments fast-growing γ-proteobacteria dominated; in the marine experiments primarily different Vibrio and Alteromonas species appeared, while in the lake water experiments species of the genera Shewanella, Aeromonas, and Rheinheimera grew. These results suggest that major, but different, γ-proteobacterial genera in both freshwater and marine environments have a preference for elevated concentrations of nutrients and easily assimilated organic carbon sources but are selectively outcompeted by α-proteobacteria in the presence of low nutrient concentrations.

The use of seawater dilution cultures for the study of bacterial growth kinetics and patterns of organic matter utilization was pioneered by Sieburth et al. (51). In subsequent studies researchers developed the technique, and it became evident that this approach was a useful experimental system for the study of marine microbial ecology (2, 20, 29). In the dilution culture approach, sterile filtered lake water or seawater is inoculated with a small amount of a mixed bacterial community, and the ensuing growth and its consequences are monitored. Since the early 1980s many aspects of globally important activities of bacterioplankton have been studied by using lake water and seawater dilution cultures. For instance, this approach has been used for studies of bacterial growth on various fractions of dissolved organic carbon (DOC) (3, 59), bacterioplankton growth efficiency (53), and bacterial mortality due to grazing and virus infection (28, 38). Critical insights into the role of marine bacterioplankton in the dissolution and regeneration of diatom silica (5), and in producing recalcitrant DOC (39, 46, 54) were obtained by using seawater dilution cultures. Measurements of the increases in bacterial biomass in seawater cultures have enabled the determination of empirical conversion factors for leucine or thymidine measurements of bacterial production needed to estimate oceanic carbon flux. Only recently has the intriguing issue of which bacteria actually proliferate in dilution cultures been addressed by molecular biology techniques (14, 16, 32).

Knowledge of the limiting nutrient for phytoplankton growth frequently aids in explaining the presence and/or abundance of specific algal groups and thereby provides insight into a major controlling factor of aquatic ecosystems. A corresponding understanding of bacterioplankton growth is a desirable goal. A bioassay frequently used to examine nutrient limitation of bacterial growth involves incubating unfiltered lake water or seawater with low concentrations of potentially limiting nutrients and monitoring the growth response in terms of changes in total bacterial numbers and production. Alternatively, the bacterial growth response can be monitored after nutrient addition to dilution cultures. By using this approach, P limitation of bacterial growth has been demonstrated in lakes (13, 37, 58). Likewise, bioassays have demonstrated the occurrence of bacterial P limitation (12, 44, 65), C limitation (10, 30), or Fe limitation (40) in several marine environments. To date, few studies have investigated the species response in such experiments. An exception is the study of Hutchins et al. (26), who monitored bacterial species composition in response to Fe addition in various Fe-limited marine areas. Significant growth of γ-proteobacteria in the Fe-supplemented preparations was observed, even though the changes in bacterioplankton were “surprisingly minor compared to the changes in phytoplankton community composition.” Despite the difficulty of distinguishing between direct and indirect effects of Fe addition on the bacterioplankton, the authors noted that the experimental approach “provided a realistic picture of likely changes” in bacterial assemblages after episodic natural Fe enrichment (26). This suggestion could also apply to the consequences of nutrient addition in areas of the sea where C, N, or P availability limits bacterial growth.

It is frequently assumed that opportunistic bacteria (notably γ-proteobacteria) will become dominant in experiments in which lake water or seawater is incubated in micro- or mesocosms (16, 18, 19). This is thought to be due mainly to the rapid growth capacity and ability to exploit elevated-nutrient conditions of this group of bacteria. Eilers et al. (16) pointed out that this characteristic of γ-proteobacteria “will confront microbiologists trying to culture as yet uncultured bacteria with fundamental problems.” Alternatively, bacterial species with distinct intrinsic growth characteristics might respond to different growth conditions, whether they are experimentally induced or not. Covert and Moran (14) recently studied the utilization of high- or low-molecular-weight fractions of dissolved organic carbon by estuarine bacteria. They found no overlap in the dominant bacterial groups that developed on each substrate, suggesting that generally occurring “weed bacteria” were not important in their experiments (14). Thus, although bacteria growing in manipulated experiments may not directly represent the phylogenetic composition of the in situ assemblages at the time of sampling, knowledge about the identity of cells responding to different treatments may still provide insights into the growth requirements of different taxa.

Recent studies have demonstrated that some bacteria able to grow in culture can be important components of bacterioplankton in natural water (23, 33, 43). At this time, knowledge concerning the particular nutritional conditions influencing the growth of specific groups or species of bacteria in lake water or seawater is sparse. Our intent was to study and compare the growth of colony-forming bacterioplankton species in dilution cultures and in nutrient addition experiments with water from both freshwater and marine environments (Lake Kinneret, the eastern Mediterranean Sea, and the northern Red Sea). In addition to determining total bacterial numbers and heterotrophic production, we enumerated and identified colony-forming bacteria on Zobell agar plates. We assumed that applying the same experimental approach to freshwater and marine samples would allow a comparison of the phylogenetic responses of bacterioplankton in the different environments. Furthermore, we hypothesized that different species of bacteria would exploit the contrasting growth conditions in treatments with or without nutrient additions.

MATERIALS AND METHODS

In situ sampling.

Lake Kinneret samples were taken on seven occasions from 18 November 1999 to 24 October 2000 (Table 1) from a depth of 5 m at station A in the center of the lake (32°82′N, 35°61′E). Lake Kinneret in northern Israel is a monomictic, mesoeutrophic, relatively large lake (area, 170 km²; maximum depth, 40 m) with an annual net primary productivity of 610 g of C m² (4). In the epilimnion the DOC, total N, and total P concentrations ranged from 3 to 4 mg liter⁻¹, from 28 to 50 μM, and from 0.5 to 1.3 μM, respectively. Lake water (10 liters) was collected in two carefully acid-rinsed 5-liter polyethylene bottles by using a 5-liter Rodhe-Ruttner sampler and was kept at the in situ temperature until it was brought to the laboratory, where it was processed within 1 to 3 h.

TABLE 1.

Summary of dilution culture and whole-water experiments performed with water from Lake Kinneret (Israel), the eastern Mediterranean Sea, and Gulf of Eilat (northern Red Sea), as well as situ temperatures and chlorophyll a concentrations

Expt	Sampling location	Date	In situ conditions		Expts performed
			Temp (°C)	Chlorophyll a concn (μg liter−1)	Dilution cultures		Whole water + nutrientsc
					Unamended controls	With nutrients
KIN1	Lake Kinneret	18 November 1999	23.0	7.7	Yes
KIN2	Lake Kinneret	28 November 1999	22.6	7.4	Yes	+Pa	Yes
KIN3	Lake Kinneret	3 January 2000	21.0	6.7	Yes		Yes
KIN4	Lake Kinneret	5 March 2000	20.0	10.0	Yes	+P, +C+Pa	Yes
KIN5	Lake Kinneret	11 June 2000	25.0	6.8	Yes		Yes
KIN6	Lake Kinneret	11 September 2000	29.0	4.1	Yesb	+P	Yes
KIN7	Lake Kinneret	24 October 2000	25.0	12.0	Yes		Yes
MED1	Mediterranean Sea	16 May 2000	21.0		Yes		Yes
MED2	Mediterranean Sea (coast)	15 October 2000	25.5		Yesb	+C+P	Yes
GULF1	Gulf of Eilat	27 March 2000	23.0		Yes		Yes
GULF2	Gulf of Eilat	24 September 2000	28.0		Yesb	+C, +C+P	Yes

^aThe species composition on Zobell agar plates was not determined.

^bCommunity DNA was collected on hybridization membranes.

^cFactorial design of C, N, and P additions.

An eastern Mediterranean offshore seawater sample was collected beyond the coastal shelf, approximately 25 km off the coast (32°81′N, 34°69′E), on 15 May 2000. Water from a depth of 5 m was collected by using a Niskin bottle and was transferred to a 20-liter polycarbonate bottle and transported to the laboratory. Samples were processed on the following day. On 15 October 2000, coastal surface water (10 liters from a depth of 0.5 m) was collected in two acid-rinsed 5-liter polyethylene bottles and kept at the in situ temperature until it was brought to the laboratory for processing within 3 h. The sampling site was located in a coastal marine reserve south of Haifa, Israel (32°80′N, 34°93′E); the water depth was 2 m.

Surface seawater from the Gulf of Eilat was collected off a pier located at the Inter-University Institute for Marine Science, Eilat, Israel, in the northern Red Sea (29°51′N, 34°94′E). The site is located on the northern part of the fringing coral reef stretching around the Sinai Peninsula. Seawater was collected in spring and in autumn (Table 1) by submerging two acid-rinsed 5-liter polyethylene bottles to a depth of 0.5 m. Samples used for nutrient addition experiments were processed within 1 h, and water used for dilution cultures was processed within 14 h after transportation at the in situ temperature to the laboratory.

Dilution cultures without nutrient addition.

We determined the growth of bacteria in unenriched duplicate or triplicate lake water and seawater dilution cultures by inoculating the natural bacterial assemblage (filtered with 0.8-μm-pore-size filters) into water that was filtered with 0.2-μm-pore-size filters and used as the growth medium. Initially, sampled water was prefiltered through a GF/F filter (Whatman). After this, for each culture replicate, approximately 900 ml of sampled water to be used as medium was filtered through a 47-mm-diameter 0.2-μm-pore-size filter (Supor-200; Gelman Sciences) at <200 mm of Hg and divided into 1-liter polycarbonate bottles (Nalgene). The inoculum was prepared by gravity filtration of sampled water through 47-mm-diameter 0.8-μm-pore-size filters (MSI) and was added to the water filtered with a 0.2-μm-pore-size filter to obtain a 5- to 20-fold dilution of the inoculum. All utensils in contact with the samples were acid rinsed with 1 M HCl and extensively washed with MilliQ water prior to use. Cultures were maintained at the in situ temperature in the dark.

Bacterial growth in the dilution cultures was monitored for 5 to 8 days until it was well into the stationary phase. Subsamples were collected daily from each culture to determine total bacterial numbers (4′,6′-diamidino-2-phenylindole [DAPI] counts) and numbers of CFU (twice daily for DAPI counts during the first 2 days of incubation). The yield of bacteria in the dilution cultures was defined as the net increase in DAPI counts from the initiation of the cultures to the maximum number reached in the stationary phase. Microscopic examination of the dilution cultures occasionally revealed growth of flagellates but only after bacteria had been in the stationary phase for 2 to 4 days. Thus, we considered the growth of bacteria in the cultures to have been essentially unaffected by predators.

Dilution cultures with nutrient addition.

On some sampling dates duplicate dilution cultures were enriched with C and P, singly or in combination (Table 1). C (glucose) and P (NaH₂PO₄) were added to final concentrations of 40 and 0.6 μM, respectively. Enrichments were chosen based on the minimal addition required to elicit a growth response in nutrient addition experiments with unfiltered lake water or seawater (see below). Nutrients were added 21 to 24 h after inoculation of the cultures in order to await the results of the experiments with the enriched unfiltered samples.

Unfiltered lake water and seawater with nutrient addition.

The effect of nutrient addition on the growth of heterotrophic bacteria was examined in unfiltered, whole-water samples. Lake water or seawater was transferred to acid-rinsed 250-ml polycarbonate bottles (Nalgene) that were thoroughly rinsed with MilliQ water and sample water. C (glucose), N (NH₄Cl), and P (NaH₂PO₄) were added to final concentrations of 40, 2, and 0.6 μM, respectively, singly and in all different combinations, in duplicate. Control bottles received no nutrients. After incubation for 24 h at the in situ temperature in the dark, bacterial production, total bacterial numbers, and numbers of CFU were determined. On three occasions (experiments KIN2, KIN5, and MED2) leucine incorporation time course experiments were performed to evaluate the response of the bacterial assemblage to the nutrient additions. In all cases measurements of bacterial production and total bacterial numbers after 3 and 15 h of incubation showed a response that was lower than but qualitatively similar to the response after 24 h (data not shown).

Bacterial heterotrophic production.

Bacterial heterotrophic production was measured by using the leucine incorporation method of Kirchman et al. (31), as modified by Smith and Azam (52); however, [¹⁴C]leucine was used instead of [³H]leucine. For each sample, triplicate aliquots (1.7 ml) and a trichloroacetic acid-killed control were incubated with 20 nM (final concentration) [¹⁴C]leucine for 1 to 2.5 h at the in situ temperature in the dark.

Enumeration of bacteria and isolation procedure.

Samples used for enumeration of total bacteria were preserved with filtered formalin (pore size, 0.2 μm; final concentration, 5%), and numbers were determined by epifluorescence microscopy by using DAPI (45) within 48 h after fixation. Cells were stained with DAPI (final concentration, 2 μg ml⁻¹) for 5 min and then filtered onto black 0.2-μm-pore-size filters (Poretics) at <200 mm of Hg. Typically, 20 microscopic fields (>500 cells) were counted. The numbers of CFU in the experiments were obtained by plating 100-μl portions of undiluted lake water or seawater samples and 100-μl portions of 10-, 100-, and 1,000-fold-diluted lake water or seawater samples on triplicate Zobell agar plates (62). The plates were incubated in the dark at 15°C until no more colonies appeared (up to 15 days). Bacterial isolates were obtained by selecting colonies with different colony morphologies; frequently, isolates with the same colony morphology were isolated from samples collected on different days but from the same experiment in order to evaluate the possibility of identifying bacteria on the basis of colony morphology. The bacterial isolates that were abundant on the agar plates were identified by their 16S ribosomal DNA (rDNA) sequences. The levels of different bacteria on the agar plates were determined by counting the numbers of colonies with the dominant colony morphologies.

PCR amplification and purification of PCR product.

The 16S rDNA sequences of 124 bacterial isolates were amplified by PCR performed with Taq polymerase (Boehringer Mannheim) from DNA preparations of cultured isolates. Bacterial 16S rDNA primers 27f (AGAGTTTGATCATGGCTCAG) and 1492r (TACGGYTACCTTGTTACGACTT) were used for amplification. The 50-μl reaction mixtures contained 50 to 500 ng of template, 10 mM total deoxynucleoside triphosphates, standard 10× Taq buffer, 25 pmol of each primer, and 1 U of Taq polymerase. The PCR amplification conditions were as follows: denaturation at 95°C for 5 min and then 35 cycles of 94°C for 30 s, 52°C for 1 min, and 72°C for 1 min and finally an extension step consisting of 72°C for 7 min. A 480 DNA thermal cycler (Perkin-Elmer) was used. The PCR product was purified by using Quantum Prep PCR Kleen spin columns (Bio-Rad).

Sequencing.

Sequencing was performed with a DYEnamic ET terminator cycle sequencing premix kit (Amersham Pharmacia Biotech Inc.) with primer 27f. The PCR amplification conditions for the sequencing reactions were as follows: 29 cycles of 96°C for 30 s, 50°C for 15 s, and 60°C for 1 min. The sequencing reaction mixture was purified by using spin columns (AutoSeq G-50; Amersham Pharmacia Biotech Inc.). 16S rDNA nucleotide sequences were determined from the purified PCR product by automated sequencing by using ABI PRISM dye terminator cycle sequencing (Perkin-Elmer). The average length of the resulting partial 16S rDNA sequences was 500 to 700 bp.

Whole-genome DNA hybridization.

In three experiments (experiments KIN6, MED2, and GULF2) bacterial community DNA was collected from the unenriched dilution cultures after entry into the stationary phase. With these samples the levels of specific bacteria dominant on agar plates from the cultures were determined by whole-genome DNA hybridization with community DNA by using the species density protocol of Pinhassi et al. (43). Briefly, the procedure was as follows. Community DNA samples were prepared by filtering 15-ml portions of water from dilution cultures onto hybridization membranes, lysing the cells with sodium hydroxide, and linking the DNA to the hybridization membranes by UV exposure. Standard curves based on known numbers of cells were prepared by using the same protocol that was used for the samples. Whole-genome DNA probes were then prepared by labeling the genomic DNA prepared from each of the different bacteria with a nick translation kit (Promega) and [α-³²P]dATP (Amersham). Each probe was hybridized to triplicate community DNA samples and to the specific standard under stringent conditions for hybridization (69°C) and washing (two 30-min washes in 2× SSC-0.5% sodium dodecyl sulfate at the hybridization temperature and two 5-min washes in 0.1× SSC at room temperature [1× SSC is 0.15 NaCl plus 0.015 M sodium citrate]). The hybridization signal was detected and quantified with a PhosphorImager (Molecular Dynamics). The abundance of the specific bacteria was obtained by relating the hybridization signal of the samples to the hybridization signal of the standard with a known number of cells.

Cross hybridization between isolates quantified by the species density protocol was determined by slot blot hybridization of extracted genomic DNA. Blotting of the DNA was done as described by the membrane manufacturer (Amersham). The DNA extracts were diluted in 10× SSC, denatured at 95°C for 5 min, transferred to ice, and then filtered onto hybridization membranes (Hybond-N; Amersham) with a slot blot apparatus (GIBCO BRL), after which the DNA was denatured, neutralized, and linked to the membrane as described above for sample preparation. Hybridization was performed as described above. As previously demonstrated (25), 16S rDNA sequence similarities around 97% between our isolates resulted in cross hybridization levels between 20 and 40%. The levels of cross hybridization fell with decreasing 16S rDNA sequence similarity. Cross hybridization was subtracted from the abundance determined for the quantified bacteria.

Nucleotide sequence accession numbers.

The 16S rDNA sequences of the isolates have been deposited in the GenBank database under accession numbers AY136072 to AY136135 (Table 2).

TABLE 2.

Bacteria isolated from dilution cultures and nutrient enrichment experiments from Lake Kinneret, the Mediterranean Sea, and the Gulf of Eilat

Isolatea	Accession no.	Taxonb	Closest relative in GenBank database (accession no.)	% Similarity	Environment where dominant (no. of experiments)c
					KIN	MED	GULF
R34	AY136126	CFB	Marine gliding bacterium UWA-1 (AB039966)	94.2	ND	+	+
R13	AY136123	CFB	Haerentibaculum mesophilum (AB032501)	97.1	ND	ND	−
M75	AY136112	CFB	Flexibacter echinicida strain F11 (AY006470)	95.0	ND	+	ND
K30	AY136072	CFB	Uncultured bacterium BA8 (AF087050)	94.4	+ (2)	ND	ND
K107	AY136085	CFB	Flavobacteriaceae strain 2 (AB024308)	95.7	+	ND	ND
R85	AY136135	α	Roseobacter gallaeciensis (Y13244)	97.4	ND	ND	−
R81	AY136134	α	Roseobacter gallaeciensis (Y13244)	99.7	ND	ND	−
M61	AY136107	α	Uncultured eubacterium HstpL28 (AF159650)	96.4	ND	+	ND
R1	AY136122	α	Slope strain D14 (AF254104)	96.6	ND	ND	−
R15	AY136124	α	α-Proteobacterium CV1010 (AF114485)	97.7	ND	−	−
R68	AY136132	α	CVSP bacterium CV1010-362 (AF114485)	96.8	ND	ND	−
M7	AY136103	α	α-Proteobacterium MBIC3951 (AB018689)	100	ND	−	ND
M6	AY136102	α	α-Proteobacterium MBIC3951 (AB018689)	96.2	ND	−	−
M8	AY136104	α	Abyssal strain AIII3 (AF254101)	96.4	ND	−	ND
M1	AY136101	α	Hydrothermal Vent strain TB66 (AF254109)	96.9	ND	−	ND
R59	AY136130	α	Marine α-proteobacterium AS-19 (AJ391181)	95.7	ND	ND	−
K136	AY136089	α	Paracoccus aminovorans (D32240)	98.4	−	ND	ND
K150	AY136092	α	Sphingomonas sp. strain S213 (AB018439)	96.4	− (4)	ND	ND
K140	AY136091	α	Sphingomonas asaccharolytica (Y09639)	98.3	−	ND	ND
K163	AY136093	α	Sphingomonas adhesiva (D16146)	96.7	−	ND	ND
K165	AY136094	α	Sphingomonas leidyi (AJ227812)	97.7	−	ND	ND
K169	AY136096	α	Sphingomonas sp. strain CFO6 (U52146)	96.3	− (2)	ND	ND
K182	AY136097	α	Sphingomonas sp. strain MBIC3020 (AB025279)	100	+	ND	ND
K84	AY136080	α	Sphingomonas sp. strain IFO 15915 (AB033949)	98.9	−	ND	ND
K48	AY136075	α	Asticcacaulis excentricus (AJ247194)	96.1	− (3)	ND	ND
K137	AY136090	α	Caulobacter sp. strain MBIC3025 (AB016980)	100	−	ND	ND
K53	AY136076	α	Caulobacter sp. strain FWC38 (AJ227774)	96.9	−	ND	ND
K166	AY136095	α	Brevundimonas sp. strain FWC04 (AJ227793)	98.3	− (2)	ND	ND
K1	AY136072	α	Brevundimonas aurantica (AJ227787)	100	− (5)	ND	ND
K131	AY136088	α	Methylobacterium extorquens (D32224)	99.8	−	ND	ND
K72	AY136077	β	Limnobacter thiooxidans (AJ289885)	99.8	−	−	ND
K109	AY136086	β	β-Proteobacterium A0637 (AF236004)	99.6	+ (2)	ND	ND
K192	AY136099	β	β-Proteobacterium MBIC3293 (AB022678)	99.5	+	ND	ND
K190	AY136098	β	Ultramicrobacterium strain 12-3 (AB008507)	93.5	+	ND	ND
M102	AY136118	γ	Alteromonas sp. strain KE10 (AB015135)	95.5	ND	+	+
M64	AY136108	γ	Alteromonas macleodii (Y18231)	99.8	ND	+	ND
M58	AY136106	γ	Uncultured γ-proteobacterium (AF114520)	95.3	ND	+	ND
M76	AY136113	γ	Alteromonas alvinellae (AF288360)	100	ND	+	ND
M100	AY136117	γ	Alteromonas macleodii (Y18228)	98.4	ND	+	ND
M87	AY136114	γ	Alteromonas alvinellae (AF288360)	99.6	ND	+ (2)	ND
R38	AY136127	γ	Uncultured Alteromonadaceae (AF513940)	99.2	ND	ND	+
R39	AY136128	γ	Vibrio alginolyticus (X74691)	99.3	ND	ND	+ (2)
M22	AY136105	γ	Vibrio splendidus biovar II (AB038030)	99.9	ND	+	+
M71	AY136110	γ	Vibrio pelagicus (AJ293802)	99.7	ND	+	+ (2)
M67	AY136109	γ	Vibrio tubiashi (X74725)	97.9	ND	+	+
R29	AY136125	γ	Vibrio pelagicus (AJ293802)	100	ND	ND	+
R47	AY136129	γ	Vibrio penaeicicida (AJ437191)	96.9	ND	ND	+
M73	AY136111	γ	Endocytic bacterium Noc2 (AF262743)	96.7	ND	+	ND
M103	AY136119	γ	Vibrio hollisae (X74707)	95.6	ND	+	ND
K197	AY136100	γ	Vibrio mimicus (X74713)	99.8	+	ND	ND
K89	AY136081	γ	Rheinheimera baltica (AJ002006)	95.2	+ (3)	ND	ND
K35	AY136074	γ	γ-Proteobacterium HTB082 (AB010842)	94.7	+ (2)	ND	ND
K103	AY136084	γ	Aeromonas veronii (AF099024)	100	+ (4)	ND	ND
K76	AY136078	γ	Aeromonas salmonicida (AF134065)	100	+	ND	ND
K80	AY136079	γ	Shewanella putrefaciens (X81623)	99.9	+ (2)	ND	ND/PICK> M91	AY136115	γ	Marinobacter arcticus (AF148811)	97.5	ND	−	ND
M104	AY136120	γ	Marinobacter sp. (AF237685)	96.4	ND	−	ND
M106	AY136121	γ	Marinobacter sp. (AF237685)	98.1	ND	−	ND
R77	AY136133	γ	Uncultured γ-proteobacterium (AY033301)	89.9	ND	ND	− (2)
M92	AY136116	γ	Oceanospirillum linum (M22365)	91.1	ND	−	ND
R65	AY136131	γ	Oceanobacter kriegii (AB006767)	94.0	ND	ND	−
K116	AY136087	γ	Pseudomonas anguilliseptica (X99541)	99.2	− (2)	ND	ND
K90	AY136082	γ	Pseudomonas pavonanceae (D84019)	99.4	+ (2)	ND	ND
K93	AY136083	γ	Acinetobacter lwoffii (X81665)	99.8	−	ND	ND

^aThe bacteria are arranged phylogentically on the basis of an unrooted consensus tree (data not shown), and major clades, roughly corresponding to separate genera or families, are separated by spaces.

^bCFB, Cytophaga-Flexibacter-Bacteroides phylum; α, α-proteobacteria; β, β-proteobacteria; γ, γ-proteobacteria.

^cKIN, Lake Kinneret; MED, Mediterranean Sea; GULF, Gulf of Eilat. +, bacterium dominant in enrichment cultures; −, bacterium dominant in unenriched dilution cultures; ND, not detected.

RESULTS

A summary of the experiments carried out in this study is shown in Table 1. A total of 298 isolates were obtained in these experiments. The 16S rDNA sequences of 124 isolates were determined in order to evaluate the phylogenetic diversity of the bacteria. The sequenced bacteria belonged to 64 tentative species that were found to be abundant on the agar plates used to determine the number of colony-forming bacteria in our experiments (Table 2).

Dilution cultures without nutrient addition.

The growth of bacterioplankton at ambient nutrient concentrations was investigated in dilution cultures grown with filtered lake water and seawater free from protozoan grazing. An example of the growth kinetics of bacteria in Lake Kinneret dilution cultures with no added nutrients is shown in Fig. 1. The yield of bacteria in these cultures varied between 2.8 × 10⁵ and 5.0 × 10⁵ cells ml⁻¹ during most of the year, with a higher yield (1.01 × 10⁶ cells ml⁻¹) recorded only in experiment KIN3 (Table 3). The average growth rate calculated from the increase in total bacterial numbers (DAPI counts) was 0.62 ± 0.27 day⁻¹ during the exponential phase, although on some dates growth appeared to be slower and not exponential. The average growth rate calculated from the increase in CFU was 1.57 ± 0.40 day⁻¹ and differed significantly from the rates calculated from DAPI counts (Student's t test; n = 17; P < 0.005). The proportion of cells forming colonies on Zobell agar plates at the end of the experiments was around 20% of the yield in five of seven experiments (range, 12 to 46%) (Table 3). In comparison, the CFU in the initial samples comprised only <0.002% of the DAPI counts. α-Proteobacteria constituted a majority of the colonies found in the lake dilution cultures during the year (Fig. 2). Sphingomonas species were found in all unenriched dilution cultures. Other frequently abundant species were Brevundimonas aurantica (strain K1) and Asticcacaulis sp. (strain K48). Although present, γ-proteobacteria like Acinetobacter sp. (strain K93), Pseudomonas anguilliseptica (strain K116), and Rheinheimera sp. (strain K89) rarely were abundant in these cultures. The somewhat unusual species composition in experiment KIN3 dilution cultures (with β- and γ-proteobacteria dominant) coincided with the highest bacterial yield of all lake samples, suggesting that there were different growth conditions on this day, perhaps because of Jordan River inflow caused by heavy rainfalls some days previously.

FIG. 1.

Increases in DAPI counts (solid symbols) and numbers of CFU (open symbols) in unenriched Lake Kinneret dilution cultures in March 2000 (experiment KIN4). The different symbols represent bacterial numbers in three replicate cultures. Total bacterial numbers were determined by DAPI staining, and numbers of CFU were obtained by spreading decimal dilutions of dilution culture water onto Zobell agar plates. The error bars indicate standard deviations (n = 20 and n = 3 for DAPI counts and CFU, respectively).

TABLE 3.

Bacterial numbers in dilution cultures from Lake Kinneret, the Mediterranean Sea, and the Gulf of Eilat

Expt	Treatment	No. of replicates	Yield (105 cells ml−1)a	CFU (105 ml−1 mean ± SD)	% CFU (% of yield)	Growth rate (day−1, mean ± SD)b
						DAPI counts	CFU
KIN1	Control	2	2.82	0.61 ± 0.05	21.6	0.83 ± 0.05	1.20 ± 0.10
KIN2	Control	3	3.37	0.41 ± 0.05	12.2	0.24 ± 0.02c	1.08 ± 0.07
	+P	2	12.31	3.94 ± 1.06	32.0	0.66 ± 0.05
KIN3	Control	3	10.10	2.10 ± 0.48	20.8	0.87 ± 0.05	1.89 ± 0.20
KIN4	Control	3	4.78	1.05 ± 0.21	22.0	0.95 ± 0.08	1.62 ± 0.02
	+P	2	5.22	1.06 ± 0.17	20.4	0.97 ± 0.08
	+C+P	2	14.66	12.02 ± 1.08	82.0	1.99 ± 0.03
KIN5	Control	2	4.52	0.86 ± 0.27	19.1	0.56 ± 0.02	1.51 ± 0.03
KIN6	Control	2	5.00	2.30 ± 0.25	46.0	0.29 ± 0.03c	2.04 ± 0.07
	+P	2	22.80	17.1 ± 1.51	74.9	0.73 ± 0.12	2.22 ± 0.09
KIN7	Control	2	3.72	0.79 ± 0.18	21.3	0.37 ± 0.05c	1.07 ± 0.09
MED1	Control	3	6.34	0.17 ± 0.03	2.7	1.33 ± 0.07	2.22 ± 0.19
MED2	Control	2	12.30	0.27 ± 0.03	2.2	1.72 ± 0.05	1.43 ± 0.11
	+C+P	2	22.30	6.15 ± 0.36	27.6	1.54 ± 0.03	2.35 ± 0.14
GULF1	Control	2	9.43	0.55 ± 0.10	5.8	2.54 ± 0.14	0.94 ± 0.10
GULF2	Control	2	4.43	0.65 ± 0.14	14.7	1.70 ± 0.07	1.46 ± 0.10
	+C	2	6.23	1.60 ± 0.14	25.7	2.06 ± 0.07	2.88 ± 0.14
	+C+P	2	18.61	7.89 ± 1.86	42.4	2.54 ± 0.03	3.32 ± 0.22

^aNet increase in number of cells.

^bThe growth rates in dilution cultures were calculated from the increases in total bacterial numbers (DAPI counts) and CFU during exponential growth.

^cGrowth was not exponential.

FIG. 2.

Bacterial species compositions on Zobell agar plates in unenriched dilution cultures and nutrient-amended cultures in Lake Kinneret (KIN), Mediterranean Sea (MED), and Gulf of Eilat (GULF) experiments. The enrichments consisted of glucose (C), ammonium (N), and phosphate (P) added to either dilution cultures or unfiltered lake water or seawater. The abundance of specific bacteria was analyzed upon entry into the stationary phase in dilution cultures and after 24 h of incubation in experiments with unfiltered water. Dil. cult., dilution culture; Not id., not identified; α-Proteob., α-proteobacteria; β-Proteob., β-proteobacteria; γ-Proteob., γ-proteobacteria.

In dilution cultures from the Mediterranean Sea and the Gulf of Eilat the yields of bacteria ranged from 4.4 × 10⁵ to 12.3 × 10⁵ cells ml⁻¹ (Table 3). The growth rates calculated from the increases in both DAPI counts and CFU ranged from 0.94 to 2.54 day⁻¹ and did not differ significantly (Student's t test). The proportions of CFU were around 2.5% and 6 to 15% of the yield in the Mediterranean Sea and Gulf of Eilat dilution cultures, respectively. In experiment MED1 (open sea surface water) the CFU were entirely dominated by five members of the Roseobacter clade. In experiment MED2 (coastal water), four different γ-proteobacteria in the Oceanospirillum clade and a β-proteobacterium, all of which formed small nondistinctive colonies that grew slowly, were most prominent (Fig. 2). In the GULF dilution cultures a majority of the bacterial CFU also belonged to the Roseobacter clade. A total of seven taxa in this group were identified, and two of these taxa were identical to isolates found in the MED1 experiment. In both GULF experiments, a novel γ-proteobacterium (strain R77), which was distantly related to the genera Marinobacter and Oceanospirillum, was also found to be abundant (Fig. 2).

Verification of CFU abundance with whole-genome DNA hybridization.

In one dilution culture experiment from each of the three aquatic environments (experiments KIN6, MED2, and GULF2) bacteria dominant in the CFU analysis were quantified by whole-genome DNA hybridization with community DNA. This was done in order to obtain culture-independent quantification of the bacteria enumerated visually from the agar plates. Hybridization revealed that the dominant bacteria in these dilution cultures accounted for 34 to 78% of the bacterial yield, although the CFU of these species accounted for only 2.2 to 37% of the yield (Fig. 3).

FIG. 3.

Yields of bacteria in dilution cultures from each of the three aquatic environments compared to the abundance of specific bacteria determined from the CFU data and by whole-genome DNA hybridization with community DNA (HYB). Not id., not identified.

Dilution cultures with nutrient addition.

Table 3 shows the results of nutrient addition to bacterioplankton grown in dilution cultures. These experiments, aimed at comparing the effects of enrichment on the growth of particular bacteria in dilution cultures and unfiltered samples, essentially indicated that primarily closely related γ-proteobacteria responded to enrichment independent of the experimental approach.

In experiments KIN2 and KIN6 addition of P increased the bacterial yield approximately fourfold compared to the yield in unenriched controls (to 1.23 × 10⁶ and 2.28 × 10⁶ cells ml⁻¹, respectively), and the growth rates increased two- to threefold. In experiment KIN4 cultures, bacteria responded only to combined addition of C and P; in these cultures the yield increased threefold and the growth rates doubled. In the dilution cultures in which nutrients stimulated bacterial yield, the proportion of colony-forming bacteria increased severalfold compared to the proportion of colony-forming bacteria in control cultures, and these organisms accounted for 32 to 82% of the yield (Table 3). The CFU that developed in experiment KIN6 dilution cultures with P addition were dominated by the γ-proteobacteria Rheinheimera sp. (strain K89) and P. anguilliseptica (strain K116) and a β-proteobacterium (strain K190), together with the α-proteobacteria Brevundimonas sp. (strain K166) and Sphingomonas sp. (strain K182) (Fig. 2).

In experiment MED2 dilution cultures, combined addition of C and P caused a doubling of the bacterial yield compared to the yield in cultures with no addition, concomitant with a 23-fold increase in the number of CFU (to 27% of the yield) (Table 3). The CFU in this experiment were dominated by different strains related to the γ-proteobacteria Alteromonas macleodii (i.e., strains M102, M87, M100) and Vibrio sp. (strain M103) (Fig. 2). In comparison, the colony-forming bacteria in the unenriched cultures were dominated by the Oceanospirillum clade.

In experiment GULF2 dilution cultures addition of C alone resulted in a slight increase in the bacterial yield, while combined addition of C and P caused a fourfold increase in the yield and a significant increase in the growth rate compared to the yield and growth rate in control cultures (Table 3). Addition of C and P to the dilution cultures resulted in a 12-fold increase in the CFU (to 42% of the yield), with dominance of the γ-proteobacteria Alteromonas sp. (strain M102), Vibrio alginolyticus (strain R39), and Vibrio tubiashi (strain M67) (Fig. 2).

Unfiltered lake water and seawater with nutrient addition.

Enrichments of unfiltered samples were prepared to investigate the growth response of the total bacterial community and the growth of specific bacterial species when nutrient availability was increased. In Lake Kinneret P alone increased bacterial production on four of six sampling dates from late spring to early winter, while a single addition of C resulted in increased bacterial growth in the early spring (Fig. 4A). In the winter, bacterial production increased only when C and P were added together. Addition of C and P and/or C, N, and P generally caused the largest increase in bacterial production, although in summer little response was observed to any nutrient addition (Fig. 4A). Concomitant with the rise in bacterial production, the average level of CFU increased to 2.3 × 10⁵ CFU ml⁻¹ in the preparations to which C, N, and P were added (Table 4). Since the level of CFU in the initial water was <3.0 × 10³ CFU ml⁻¹ during the year, the cultivable fraction increased from <0.002% to 3 to 10% of the DAPI counts upon incubation with C, N, and P. In these nutrient addition experiments mainly γ- and β-proteobacteria, as well as Cytophaga relatives, grew (Fig. 2). In particular, the γ-proteobacteria Aeromonas veronii (strain K103) and Rheinheimera sp. (strains K35 and K89) were dominant throughout the year. Other repeatedly occurring species were Shewanella putrefaciens (strain K80), Pseudomonas pavonanceae (strain K90), a β-proteobacterium (strain K109), and a Cytophaga sp. strain (strain K30). The α-proteobacteria B. aurantica (strain K1) and Caulobacter sp. (strain K137) were significant in nutrient addition experiments only in experiments KIN2 and KIN6, respectively (Fig. 2).

FIG. 4.

Comparison of leucine incorporation rates in situ and 24 h after addition of nutrients to unfiltered surface water from Lake Kinneret, the Mediterranean Sea, and the Gulf of Eilat. Treatments included addition of glucose (C), ammonia (N), and phosphate (P), singly and in different combinations, as well as an unenriched control (0). The error bars indicate standard deviations for the pooled measurements of triplicate subsamples from each duplicate treatment (n = 6). Note the interrupted scale of the y axis in the panel for experiment MED1.

TABLE 4.

Comparison of total bacterial numbers (DAPI counts) and number of colony-forming bacteria (CFU) in whole-water nutrient addition experiments for Lake Kinneret, the Mediterranean Sea, and the Gulf of Eilat

Expt	Total bacterial no. (106 cells ml−1, mean ± SD)			Colony-forming bacteriaa
	In situ	Controla	With C+N+Pa	With C+N+P (106 CFU ml−1, mean ± SD)	% of DAPI counts (% of DAPI net increase)
KIN1	2.81 ± 0.42
KIN2	2.68 ± 0.30	2.55 ± 0.41	3.22 ± 0.76	0.20 ± 0.05	6 (37)
KIN3	2.65 ± 0.45	2.75 ± 0.38	3.39 ± 0.61	0.31 ± 0.05	9 (42)
KIN4	2.12 ± 0.29	2.14 ± 0.35	2.57 ± 0.53	0.27 ± 0.08	10 (59)
KIN5	2.07 ± 0.31	1.89 ± 0.21	2.44 ± 0.36	0.10 ± 0.02	4 (26)
KIN6	12.30 ± 2.36	12.52 ± 2.27	14.13 ± 3.33	0.44 ± 0.16	3 (24)
KIN7	3.28 ± 0.45	3.14 ± 0.47	3.93 ± 0.72	0.29 ± 0.06	7 (44)
MED1	0.93 ± 0.18	1.04 ± 0.16	1.49 ± 0.28	0.46 ± 0.07	31 (83)
MED2	3.48 ± 0.38	3.66 ± 0.62	4.73 ± 0.71	0.64 ± 0.05	13 (51)
GULF1	0.53 ± 0.09	0.62 ± 0.12	1.31 ± 0.24	0.64 ± 0.14	49 (82)
GULF2	2.11 ± 0.31	2.22 ± 0.33	2.79 ± 0.42	0.26 ± 0.03	9 (38)

^aAfter 24 h of incubation at the in situ temperature.

On both Mediterranean Sea sampling dates combined addition of C and P was required to trigger an increase in bacterial production (Fig. 4B). Bacterial production further increased severalfold in preparations treated with C, N, and P. The level of CFU initially was <800 CFU ml⁻¹ but rose to 4.6 × 10⁵ and 6.4 × 10⁵ CFU ml⁻¹ in preparations treated with C, N, and P in experiments MED1 and MED2, respectively, comprising up to 31% of the DAPI counts (Table 4). In experiment MED1 the entire increase in CFU could be attributed to the growth of Vibrio splendidus (strain M22) (Fig. 2). In experiment MED2 three different Vibrio species (V. pelagicus, V. tubiashi, and a Vibrio sp.) and a Flexibacter sp. strain (strain M75) accounted for the increase in CFU. On 22 May a second addition experiment was performed with seawater originally collected for experiment MED1. Again, addition of C, N, and P triggered the largest increase in bacterial production (data not shown), and the CFU were dominated by different species affiliated with Alteromonas macleodii (strains M64, M58, and M87) and a Roseobacter sp. strain (strain M61) (Fig. 2).

In Gulf of Eilat waters, C was the only nutrient that caused a significant increase in bacterial production when it was added alone (Fig. 4C). In experiment GULF1 bacterial production did not increase further with combined additions of C, N, and/or P. On 30 March a second nutrient addition experiment was carried out after 3 days of incubation of the sampled water at the in situ temperature, and a similar pattern of response to nutrients was observed (data not shown). In experiment GULF2 combined additions of C and P and C, N, and P further increased production (Fig. 4C). Initially, the CFU in these experiments accounted for <0.002% of the DAPI counts. After the addition of C, N, and P, the level of CFU increased by more than 2 orders of magnitude, and the CFU accounted for 48.8 and 9.3% of the DAPI counts after 24 h of incubation in experiments GULF1 and GULF2, respectively (Table 4). A total of five Vibrio species together accounted for the great majority of the CFU in the GULF cultures (as shown in Table 2, three of these species were also dominant in the MED experiments). V. pelagicus (strain M71) and V. alginolyticus (strain R39) were abundant in both GULF experiments (Fig. 2). Also present were a novel Cytophaga relative (strain R34) and two species related to Alteromonas (strains R38 and M102).

DISCUSSION

The significant increases in bacterial growth rates and yields in response to nutrient addition that we observed in experiments with both dilution cultures and unfiltered water indicated that the bacterioplankton in nearly all our samples were nutrient limited. This may seem to be an obvious observation, since nutrient limitation could be expected to be a frequent condition in most natural waters. Nevertheless, bacterioplankton are at times little limited by nutrient availability, whereas the impact of water temperature is usually considerable, whether nutrients are abundant or not (50, 64). It might be argued that the growth responses which we observed in this series of experiments were merely a consequence of sample incubation in small containers (i.e., that there were nonspecific confinement effects). However, since there were no major differences between bacterial production in situ, as determined prior to the experiments, and bacterial production in controls after 24 h of incubation, we concluded that the effects of confinement were negligible compared to the effects of the added nutrients.

In Lake Kinneret, we found considerable temporal variability in the amplitude of the bacterial growth response to nutrient addition, although at most times P appeared to be limiting. These results are in line with a number of studies demonstrating frequent P limitation and occasional C limitation of bacterioplankton in lakes (11, 13, 37, 58). In our Mediterranean Sea samples combined addition of C and P was required to elicit an increase in bacterial production compared to the production in the controls, suggesting that there was colimitation by C and P at the time of sampling. Zohary and Robarts concluded that P is the primary limiting nutrient for bacterial and phytoplankton growth over large areas of the eastern Mediterranean Sea in the winter (63). Likewise, P limitation has been reported frequently in northwestern Mediterranean Sea surface waters in the summer (48, 57, 65), while C or N at times limits bacterial production in the deep chlorophyll maximum (48). In the coastal Gulf of Eilat, we found that bacterioplankton growth was limited by the availability of organic carbon both in the spring and in the autumn. This result apparently contrasts with the observation of very low phosphate concentrations and high alkaline phosphatase activities in the picoplankton size fraction, suggesting that there is P limitation in the surface waters of central areas of the Gulf of Eilat (35). However, for the coastal areas of the gulf, Richter et al. recently showed that high levels of N and P regeneration are associated with crevice-dwelling sponges in coral reefs, suggesting an explanation for our observed C limitation of bacterioplankton residing close to the reefs (47).

In our dilution culture and nutrient addition experiments the number of CFU and the proportion of DAPI counts that the CFU accounted for greatly increased during incubation. In the original water samples, the CFU were typically <0.002% of the total bacterial numbers counted by DAPI staining, whereas after incubation the CFU accounted for up to 20 to 80% of the DAPI counts. The greatest increase in CFU was found in experiments with added nutrients. In three dilution culture experiments (experiments KIN6, MED2, and GULF2) whole-genome DNA hybridization was used to verify that it was feasible to quantify bacteria that appeared to be abundant on the agar plates by counting colonies with distinct morphologies on the plates. The results of these experiments showed that the isolated bacteria accounted for larger proportions of the experimental bacterial communities than the proportions actually indicated by the numbers found on the agar plates. Our results are consistent with the observation of Kisand et al. (32) that in dilution cultures, the CFU underestimated the abundance of specific bacteria. In the dilution cultures, growth of new cells was evident, and it was reasonable to assume that the increase in colony-forming bacteria was due to growth of these new cells. In the enrichment experiments with unfiltered water, on the other hand, the increase in bacterial numbers within 24 h was usually small compared to the DAPI counts measured at the start of the experiments. Although it was not explicitly studied here, we assumed that the increase in CFU in these experiments was also a consequence of the growth of bacteria able to form colonies on solid media rather than activation of dormant cells in the original community. This assumption was based on the following findings. First, we are aware of no study that has unambiguously demonstrated that gram-negative bacteria display dormancy, although they are liable to go through a number of physiological states ranging from highly active to death when they are exposed to severe starvation (27). Second, in enrichment cultures similar to our cultures, monitored by fluorescence in situ hybridization, it was demonstrated that an increase in cell number was due to active cell division (16). Third, it has been shown that in cultures activation of supposedly dormant cells was in fact due to growth of residual active cells in the cultures (6, 7). As shown in Table 4, the colony-forming bacteria in most of our enrichment experiments with unfiltered water accounted for 3 to 10% of the DAPI counts after 24 h of incubation but on average comprised 48% of the newly grown cells. Recent experiments demonstrated that these percentages increased considerably if the incubation time was extended (Pinhassi, unpublished data). Regardless of the exact proportions, the substantial contribution of colony-forming bacteria to the bacterial communities in our experiments warrants an analysis of the phylogenetic identity of these bacteria.

In most unenriched lake water and seawater dilution cultures the cultivable fraction of the bacterial community became dominated by α-proteobacteria that formed nondistinctive colonies (Fig. 2). However, at the species or genus level no phylogenetic overlap between the freshwater and marine experiments was found (Table 2). The most prominent feature of the marine dilution cultures was the abundance of bacteria affiliated with the Roseobacter clade. Wide distribution of members of the Roseobacter clade in seawater has been demonstrated by a variety of molecular techniques (23, 33, 49), showing that these organisms are an important component of marine bacterioplankton (21). Consequently, significant growth of Roseobacter sp. has been recorded in unenriched seawater cultures or nutrient-depleted medium in several marine areas (16, 22, 23, 36).

In the dilution cultures from Lake Kinneret, we found several members of the genus Sphingomonas and the genera Caulobacter, Brevundimonas, and Asticcacaulis. The genus Sphingomonas is widespread in both marine and freshwater environments (56). Sphingomonas alaskensis is an intensively studied species, which was originally isolated by the dilution to extinction approach and was thereafter considered a representative of marine bacterioplankton (9, 60). This bacterium has a high-affinity nutrient uptake system, and its low number of ribosomes appears to be correlated with its low growth rate and the presence of a single rRNA operon copy (15, 17). Notably, Sphingomonas species were dominant in bacterioplankton in the northern Baltic Sea under stratified summer conditions, when nutrient availability limited bacterial growth (42). Regarding bacteria in the genera Caulobacter, Brevundimonas, and Asticcacaulis, it is generally thought that their prosthecate morphology and complicated cell cycle are consistent with the physiological properties of oligotrophic bacteria with a tolerance for prolonged nutrient scarcity (1). Considering the broad range of diversity of bacteria previously isolated from the aquatic environment, the response of a limited set of α-proteobacteria in the dilution cultures without nutrient addition was a striking observation. Based on our observations and what is currently known about the growth requirements of some of these bacteria, we suggest that these related genera (i.e., Roseobacter and Sphingomonas and Caulobacter relatives), although specific to freshwater or seawater, share some aspects of physiological adaptation to growth at low ambient nutrient concentrations.

In experiments in which nutrient addition stimulated bacterial growth, the cultivable fraction of the bacterial community rapidly became dominated by fast-growing γ-proteobacteria that formed conspicuous colonies on the agar plates. This result was consistent for all three aquatic environments studied, although in most cases a contribution by Cytophaga relatives and β-proteobacteria was also evident. As shown in Table 2, mostly different Vibrio and Alteromonas species responded to addition of nutrients in the marine experiments, while in the lake water experiments species belonging to the genera Aeromonas, Shewanella, and Rheinheimera became dominant. These results suggest that among the γ-proteobacteria there is a definite phylogenetic distinction between genera that respond to elevated nutrient availability in marine and freshwater systems.

Several studies have indicated that some taxa of the γ-proteobacteria may represent bacteria with an opportunistic life strategy. In marine waters Vibrio and Alteromonas species have been repeatedly found in nutrient-enriched seawater incubations (33, 34, 36). In experiments with water from the North Sea, Eilers et al. (16) demonstrated selective growth of Vibrio and Alteromonas species in dilution cultures. In particular, addition of carbon compounds selectively stimulated the growth of Vibrio species. These authors concluded that the potential for high growth rates and the ability to maintain large amounts of ribosomes even during starvation could explain the dominance of Vibrio and Alteromonas species upon incubation (16).

The γ-proteobacteria that grew in nutrient-enriched cultures from Lake Kinneret seemed to share a preference for high-nutrient conditions with their marine counterparts. Weinbauer and Höfle (61) demonstrated that in Lake Plußee there was a high level of Aeromonas hydrophila associated with the most active microbial compartment in the metalimnion. These authors concluded that this bacterium is an opportunist which is able to exploit favorable growth conditions and has a capacity for high growth rates when it is supplied with easily degradable DOC, has an ability to utilize single carbohydrates, and has versatile nutritional and enzymatic characteristics (61). Similarly, S. putrefaciens is a fast-growing and metabolically versatile species that forms conspicuous colonies on agar plates. This species has been found in the Baltic Sea, where it is thought to be a potentially important mediator in N transformations (8). Thus, it appears that the capacity for fast growth in response to abundant inorganic nutrients and easily assimilated carbon sources is a conserved trait among some important lineages of γ-proteobacteria. Furthermore, we suggest that closely related but distinct genera of the γ-proteobacteria are ecological equivalents, representing fast-growing opportunists, in freshwater and marine environments.

The species composition of the bacteria that developed in experiment MED2 dilution cultures with coastal water differed markedly from the species compositions in the other experiments in that all small colorless colonies in this experiment were affiliated with the γ-proteobacterial Oceanospirillum clade (which was present to a lesser extent also in the GULF cultures) rather than with α-proteobacteria. González and Whitman recently highlighted the diverse yet monophyletic nature of the Oceanospirillum clade, stressing that little is known about the ecology of these bacteria (24). After comparing the growth kinetics of two Pseudoalteromonas and Oceanospirillum species in a relatively low-nutrient synthetic medium, Pernthaler et al. concluded that Oceanospirillum sp. grew much more “efficiently” (41), possibly signifying that it could maintain growth at substrate concentrations far lower than those favorable for Pseudoalteromonas sp. If these observations are true, it appears that phylogenetic divergence among the cultured γ-proteobacteria also reflects significant differences in physiology of these bacteria, which range from the truly opportunistic Vibrio-Alteromonas-Shewanella lineage to the more ambient Oceanospirillum clade.

Our findings of distinct bacterial genera in different experiments are in agreement with a few previous reports (14, 32). These studies show that important information can be obtained by analysis of the growth of phylogenetically different bacteria in response to different experimental treatments by using the dilution culture approach. However, several studies have emphasized that differential filtration and confinement effects may result in the growth of fast-growing opportunists rather than the growth of specific bacteria responding to the actual treatment (16, 18, 19). Judging by the relatively high growth yields, we noted that these studies were performed in eutrophic waters in which dominant bacterial growth of similar or identical weed species may be expected irrespective of supplements. For example, using culture-independent techniques to examine the bacterial species composition in eutrophic Plymouth Sound water, Fuchs et al. (19) found dominant growth of mainly γ-proteobacteria and to a lesser degree α-proteobacteria in a series of dilution culture experiments. Thus, a prerequisite for using dilution cultures to evaluate the role of nutrient availability in structuring bacterioplankton species composition may be that the original water must not be already nutrient replete.

An important concern in the present study was the reliability of colony morphology as a discriminating character for quantifying specific bacterial taxa on Zobell agar plates in the lake water and seawater incubation experiments. With natural samples this counting strategy is certainly not feasible because as many as 25 more or less similar colony morphologies may be observed on each plate (unpublished observations). Furthermore, a plate-based quantitative estimate of colony-forming bacteria is of little value since <0.1% of the total bacterial numbers are usually recovered from aquatic environments. However, during incubation of our experimental samples the proportion of colony-forming bacteria increased significantly (Tables 3 and 4). Concomitantly, the visual diversity was reduced compared to that of the initial samples, and only three to seven distinguishable colony morphologies were abundant on the plates (in an extreme case only one distinguishable colony morphology was abundant). Still, there have been reports both of variation of species identity for the same colony morphology and of consistency of species identity with colony morphology (32, 55), raising doubts about the reliability of colony morphology as a discriminating character. In our hands, sequencing of duplicate isolates from the same sample (20 pairs) invariably demonstrated that colonies with the same morphology had identical or nearly identical 16S rDNA (>99.5% sequence similarity). Interestingly, sequencing of isolates with the same colony morphology collected at different times of the year in Lake Kinneret experiments also indicated that they belonged to the same species (17 cases). Thus, we are reasonably confident that in controlled experiments, where diversity is reduced compared to the diversity in in situ samples, colonies with the same morphology belong to the same species.

Only a few studies have thoroughly investigated the growth of specific bacterioplankton in seawater in response to a manipulated supply of inorganic nutrients and dissolved organic carbon (14, 16, 32), and we are unaware of similar studies in lake water. Here we report marked differences in the amplitude of the bacterial growth response between experiments with and without added nutrients. Consequently, the capacity for fast growth in response to elevated nutrient concentrations resulted in the dominance of opportunistic γ-proteobacteria in nutrient-enriched cultures. However, these bacteria were relatively unsuccessful compared to α-proteobacteria in control cultures with low ambient nutrient concentrations. Based on our experience, we suggest that detailed examination of bacterial growth responses to nutrient additions in lake water or seawater cultures is a useful approach, not only for analysis of important biogeochemical processes in aquatic environments (5, 39) but also for determining the responses of different bacterial species to various nutrients and organic substrates.