Effects of naphthalene on microbial community composition in the Delaware estuary


  • Editor: Julian Marchesi

  • Present address: Dawn M. Castle, University of Tennessee at Chattanooga, Biological and Environmental Sciences, 615 McCallie Avenue, Chattanooga, TN 37403, USA.

Correspondence: David L. Kirchman, College of Marine Studies, University of Delaware, Lewes, DE 19958, USA. Tel.: +1 302 645 4375; fax: +1 302 645 4028; e-mail: kirchman@udel.edu


The effects of naphthalene on microbial communities in the bottom boundary layer of the Delaware Bay estuary were investigated in microcosms using denaturing gradient gel electrophoresis (DGGE) and fluorescent in situ hybridization (FISH) with oligonucleotide probes. Three days after the addition of naphthalene, rates of bacterial production and naphthalene mineralization were higher than in no-addition controls and than in cases where glucose was added. Analyses using both DGGE and FISH indicated that the bacterial community changed in response to the addition of naphthalene. FISH data indicated that a few major phylogenetic groups increased in response to the glucose addition and especially to the naphthalene addition. DGGE also demonstrated differences in community composition among treatments, with four phylotypes being unique to naphthalene-amended treatments and three of these having 16S rRNA genes similar to known hydrocarbon degraders. The bacterial community in the naphthalene-amended treatment was distinct from the communities in the glucose-amended treatment and in the no-addition control. These data suggest that polycyclic aromatic hydrocarbons may have large effects on microbial community structure in estuaries and probably on microbially mediated biogeochemical processes.


Contamination by organic pollutants is a problem in many coastal and estuarine waters adjacent to urban areas. Polycyclic aromatic hydrocarbons (PAHs) are of particular concern because of their continuous release, persistence in the environment, and the well-known toxic effects on bottom-dwelling organisms, such as fish, mollusks, and invertebrates (Mueller et al., 1999; Preston, 2002). However, the effects of organic pollutants, specifically PAHs, on natural microbial communities are less clear.

Microbial degradation is the primary route for the breakdown of PAHs in marine environments, and many bacteria capable of degrading PAHs have been isolated and identified (Geiselbrecht et al., 1996; Kasai et al., 2002; Zocca et al., 2004). Numerous studies have investigated the genes and biochemical pathways involved in PAH degradation (Eaton & Chapman 1992; Habe & Omori, 2003). These studies have provided us with molecular tools (Fleming et al., 1993; Hamann et al., 1999) and information about which bacteria may be involved in the degradation of pollutants in the environment. However, the culturable bacteria examined by previous studies are not necessarily the most active or numerous members of microbial communities in natural environments. In addition, it is difficult to extrapolate from laboratory experiments with cultured bacteria to natural environments with complex microbial communities.

Direct detection of 16S rRNA genes by various molecular techniques avoids the biases associated with culturing techniques and therefore provides a more complete analysis of bacterial communities in natural environments. One common molecular technique is denaturing gradient gel electrophoresis (DGGE) of 16S rRNA genes amplified by PCR (Muyzer et al., 1993). Several studies have used this technique to investigate the effects of hydrocarbons on natural communities. Kasai et al. (2001) found sequences similar to known hydrocarbon degraders in oil-paste retrieved from the site of a marine oil spill. In contrast, Chang et al. (2000) did not find any bacteria related to known hydrocarbon degraders in marine sediments exposed to oil. Two studies used DGGE to investigate the response of marine sedimentary microbial communities to experimental oil spills (Macnaughton et al., 1999; Ogino et al., 2001). Macnaughton et al. (1999) reported an increase in two major groups, Alphaproteobacteria and Cytophaga-Flexibacter-Bacteroides, while Ogino et al. (2001) observed the development of a single dominant phylotype of Gammaproteobacteria after the oil addition. No consistent relationship between PAH exposure and community composition has emerged from studies using this approach.

Detection of small subunit rRNA by fluorescence in situ hybridization (FISH) avoids some of the problems of PCR and DGGE (Amann et al., 1995). Few studies have used quantitative methods such as FISH to investigate the response of microbial communities to pollutants. Using FISH, Syutsubo et al. (2001) and Rabus et al. (1996) found that the abundance of members of the Alcanivorax-Fundibacter group (Gammaproteobacteria) and Azoarcus-Thauera group (Betaproteobacteria), respectively, increased in oil-amended enrichment cultures. These limited studies suggest that no one particular phylogenetic group is responsible for the degradation of petroleum.

In this study, we used both DGGE and FISH to investigate the effects of PAHs on natural microbial communities in the bottom boundary (nepheloid) layer of an estuary. The nepheloid layer is enriched in PAHs, compared with the water column, and may transport contaminants throughout the estuary (Pohlman et al., 2002). We chose to focus on naphthalene because it is ubiquitous in polluted waters (Kastner et al., 1994), it is the best studied of the PAHs (Ahn et al., 1999), and it is a good indicator of fresh petroleum hydrocarbon sources (Arzayus et al., 2001). We found that naphthalene affected microbial communities and that several of the phylotypes in the naphthalene-amended treatments were similar to known PAH degraders.

Materials and methods

Sample collection and experimental design

Water was collected from the Delaware Bay estuary near Philadelphia, PA (USA) in October 1999. The inoculum for the experiment was collected from the bottom boundary layer, where naphthalene and other hydrocarbon contamination is historically high (up to 48 μg g−1) (Pohlman et al., 2002; Boyd et al., 2005). Surface water was collected upstream, where hydrocarbon contamination is low (2–10μg L−1). To minimize the effects of grazers, 200 mL of whole nepheloid material was added to 1.8 L of filtered (0.2 μm pore diameter) surface water.

Diluted nepheloid material was amended with naphthalene or glucose to a final concentration of 50 μM-C. Amended treatments and an unamended control were incubated in 2 L bottles at the in situ temperature (15°C) with shaking in the dark. Subsamples for bacterial production, culturing, and naphthalene mineralization were processed immediately. Samples for DNA extractions, bacterial abundance, and in situ hybridization were immediately frozen and stored at −20°C.

Bacterial production and naphthalene mineralization

Bacterial production was estimated by 3H-leucine incorporation (Kirchman 2001). Triplicate 1.5 mL samples with 20 nM 3H-leucine were incubated for 1 h at the in situ temperature. Following cold trichloroacetic acid (TCA) extraction, incorporated leucine was collected by centrifugation and measured by liquid scintillation counting. Killed controls (5% TCA) were incubated in parallel. Bacterial abundance was estimated by epifluorescence microscopy using 4′,6′-diamidino-2-phenylindole (DAPI) and counted using a semiautomated image analysis system (Cottrell & Kirchman, 2003). In order to compare with naphthalene mineralization, leucine incorporation rates were converted to biomass production by assuming 1.5 kg C mol−1 of incorporated leucine.

Naphthalene mineralization was estimated by adding UL-14C-naphthalene (18.6 mCi mmol−1) to 10 mL nepheloid material for a final concentration of 500 ng g−1. Isotope dilution was calculated from the ambient naphthalene concentration at the station used for dilution (glucose-amended and control treatments) or the added concentration of naphthalene (Boyd et al., 2005). Samples were incubated for up to 24 h at in situ temperature and then stopped by acidification, which also forced 14CO2 into the headspace. The evolved 14CO2 was captured on NaOH-soaked filter paper suspended in the headspace of the incubation chamber.

A spray plate assay was used to assess the abundance of culturable PAH-degrading bacteria (Kiyohara et al., 1982). Inocula from experimental treatments were grown on Bushnell–Haas agar plates and sprayed with 100 mg mL−1 phenanthrene dissolved in methanol. Phenanthrene was used because it is less volatile than naphthalene and is a good indicator of naphthalene degradation potential (Hayes et al., 1999). Plates were incubated at room temperature for up to 3 weeks and were observed daily. Colonies producing clearing zones were scored positive for degradation potential.

DNA fingerprinting by DGGE

DNA was extracted using a bead-beating method to ensure separation of particle-associated bacteria. Samples (5 mL) were filtered onto 0.2 μm polycarbonate filters and placed in 2 mL screw-cap vials with 1.5 g of 0.1 mm zirconium silica beads, 250 μL 10% SDS, 500 μL 120 mM sodium phosphate buffer (pH 8.0), and homogenized in a bead beater. Two phenol:chloroform : isoamyl alcohol (25 : 24 : 1) extractions were followed by one chlorform : isoamyl alcohol (24 : 1) extraction. Sodium acetate (0.3 M) and ice-cold isopropanol were used to precipitate DNA, which was then purified with a Geneclean kit (Qbiogene, Carlsbad, CA) before PCR amplification.

The 16S rRNA gene was amplified from 25 ng of DNA in a PCR reaction with primers 338f with a 40 bp GC clamp on the 5′ end (CTCCTACGGGAGGCAGCAG) and 517r (ATTACCGCGGCTGCTGG) (Muyzer et al., 1993). Each 50 μL reaction contained 0.1 μM of each primer, 1.25 U Taqstart (Clontech, Mountain View, CA), 2.5 U taq polymerase (Stratagene, La Jolla, CA), Taq PCR buffer, 2 mM MgCl2 and 200 μM dNTPs. A hot-start PCR (MJ Designs PTC 100 thermocycler, Watertown, MA) was performed at 95°C for 5 min followed by a touchdown procedure. The annealing temperature began at 65°C and decreased by 0.5°C every cycle until it reached 55°C; then five additional cycles were carried out at 55°C followed by primer extension for five minutes at 72°C. Amplification products were checked by agarose gel electrophoresis. Three PCR reactions were pooled, concentrated with a Microcon filter concentrator (Millipore, Bedford, MA), and the DNA was quantified spectrophotometrically.

Denaturing gradient gel electrophoresis was performed at 80 V for 5 h at 60°C with a D-Code system (Bio-Rad Laboratories, Hercules, CA). At least 1500 ng of PCR product was loaded on an 8% (weight in volume) polyacrylamide gel with a denaturant gradient from 25% to 70% (100% denaturant is 7 M urea and 40% deionized formamide). Subsequently, bands were visualized by silver staining (Pharmacia Biotech, Uppsala, Sweden). The gel image was analysed with Phoretix 1D Advanced software (Durham, NC). An unweighted pair group method with arithmetic means (UPGMA) dendrogram was generated from a similarity matrix based on common band positions between lanes.

Dominant bands were stabbed with a sterile pipette tip and eluted for 20 min in sterile water at 90°C, reamplified, and subjected to DGGE to confirm the position and the presence of a single band. When necessary, bands were stabbed again and the elution process was repeated. The clean amplification product, which was about 180 bp, was sequenced on an ABI PRISM 310 (Perkin Elmer, Boston, MA) with both 358f and 517r primers using a Big Dye terminator reaction kit. BLAST searches (http://www.ncbi.nlm.nih.gov/BLAST/) were performed on all sequences to determine phylogenetic relationships.


The abundances of four major phylogenetic groups were determined by FISH with CY3-labelled oligonucleotide probes. Subsamples for FISH were thawed and fixed overnight with fresh paraformaldehyde (2% final concentration), and then sonicated for 15 s and centrifuged to remove particles that interfered with FISH analysis. The appropriate volume to obtain 3 × 107 cells per filter was filtered through 0.2 μm polycarbonate filters, rinsed three times with 0.2 μm filtered Milli-Q water, and stored at −20°C, if necessary. If not used immediately, filters were soaked overnight in 2% paraformaldehyde and rinsed just before hybridization. All filters were exposed to white light for 1 h before hybridization. The re-fixing and light treatments increased the number of bacteria detectable by FISH (unpublished results).

Domain and group-specific probes used were Eub338 for eubacteria (Amann et al., 1990), Alf968 for Alphaproteobacteria, Bet42a for Betaproteobacteria (Manz et al., 1992), Gam42a for Gammaproteobacteria (Manz et al., 1992), CF319a for the Cytophaga-Flavobacterium cluster (Manz et al., 1996), and a negative control probe for nonspecific binding (Karner & Fuhrman, 1997), which was 3% of the total abundance on average. Hybridization with the Bet42a and Gam42a probes included unlabelled competitor probes (Manz et al., 1992). For each probe a filter piece was placed on a Parafilm-covered glass slide and overlaid with 30 μL hybridization solution with 2.5 ng μL−1 final concentration of the probe. Hybridization solutions contained 0.9 M NaCl, 20 mM Tris-HCl (pH 7.4), 0.01% SDS, and the optimum concentration of formamide (Castle & Kirchman, 2004). Filters were incubated in sealed chambers at 42°C for 8–12 h. After hybridization, filters were washed for 20 min at 48°C in a wash solution (20 mM Tris-HCl pH 7.4, 5 mM EDTA, 0.01% SDS, and the appropriate concentration of NaCl). Rinsed and dried filter pieces were mounted with 4 : 1 Citifluor : Vectashield with 2 μg mL−1 DAPI in the mount. Samples were analysed using semiautomated image analysis on an Olympus Provis AX70 microscope (Lake Success, NY) with ImagePro (Media Cybernetics, Silver Spring, MD) software (Cottrell & Kirchman, 2003).

Accession numbers

All sequences obtained by this study have been deposited in Genbank under accession numbers DQ174 506−DQ174 515.


Bacterial production and mineralization of naphthalene

We investigated the effects of naphthalene additions on natural bacterial communities from the Delaware estuary. Bacterial production, as measured by leucine incorporation, was greatly stimulated in naphthalene-amended treatments, but only slightly stimulated when glucose was added relative to controls. Within 3 days, incorporation rates in the naphthalene-amended treatment were threefold greater than the control, and were 4.5-fold greater by 14 days (Fig. 1a). As expected, glucose additions stimulated bacterial production relative to the control, but to a lesser extent than naphthalene additions. On days 3 to 10, rates in the glucose-amended treatment were two- to threefold greater than in the control, but by day 14 they were not significantly different from the control. Production rates in control incubations varied relatively little over the entire incubation (Fig. 1a).

Figure 1.

 Bacterial activity after the addition of naphthalene or glucose. (a) Rates of 3H-leucine incorporation, a measure of bacterial production. (b) Mineralization of 14C-naphthalene. Data represent average values from triplicate incubations. Error bars indicate standard deviation and are omitted if smaller than symbols.

Naphthalene mineralization rates were significantly higher in the naphthalene-amended treatment than in the control and in glucose-amended treatments from day 3 to the end of the experiment (Fig. 1b). Rates in the naphthalene-amended treatment increased from nondetectable to a maximum of 105 μg C L−1 day−1 within 3 days. After day 3, mineralization rates in the naphthalene-amended treatment decreased (18–57 μg C L−1 day−1), but remained greater than rates in the control and glucose-amended treatments. Naphthalene mineralization was either not detectable or low (day 3) in the control treatment, while mineralization rates in the glucose incubation were undetectable throughout the experiment. In the naphthalene-amended treatment, the ratio of naphthalene mineralization to bacterial production was 0.5- to fivefold higher than the ratios in the glucose-amended and control treatments (Fig. 2), indicating that higher mineralization rates were not solely a result of increased bacterial production.

Figure 2.

 Ratio of naphthalene mineralization to bacterial production. The ratio was calculated from mineralization and production data (μgC L−1 day−1) at each time-point.

PAH-degrading culturable bacteria

Bacteria were isolated from the treatments and tested for the potential to degrade PAHs. Bacteria isolated from naphthalene- and glucose-amended treatments were positive for phenanthrene degradation by day 3, while no phenanthrene-degrading bacteria were isolated from the control treatment throughout the experiment (data not shown). On days 3 and 10, up to 40 cells mL−1 isolated from the glucose-amended treatment were able to degrade phenanthrene, while more than 1.0 × 103 cells mL−1 isolated from the naphthalene-amended treatment degraded phenanthrene. By the last sampling day, only 10 cells mL−1 isolated from the glucose-amended treatment and more than 1.5 × 103 cells mL−1 from the naphthalene-amended treatment exhibited the potential to degrade PAHs.

DGGE analysis of PCR-amplified 16S rRNA gene fragments

Denaturing gradient gel electrophoresis analysis indicated that the microbial community changed over time and between treatments (Fig. 3). The naphthalene-amended treatment diverged from the glucose-amended treatment and control within 3 days, and these differences persisted to the end of the experiment (Fig. 4). In the naphthalene-amended treatment, there were eight dominant bands, compared with four and three in the glucose-amended treatment and control, respectively. Three of the bands that first appeared in the naphthalene-amended treatment on day 3 (bands 3, 4 and 5) were also present on day 14. There were four dominant bands on day 3 in the glucose-amended sample, two of which were also present on day 14 in the glucose-amended treatment. The control treatment usually had the fewest dominant bands (3 to 7 bands). On day 3 there was no apparent difference between the control and inoculum, but by day 10 only one of the original bands was visible.

Figure 3.

 Denaturing gradient gel electrophoresis analysis of bacterial communities. Lanes represented are inoculum (I), control (C), glucose (G), and naphthalene (N). Dominant bands (1–10) were excised and sequenced.

Figure 4.

 UPGMA dendrogram of bacterial communities revealed by the denaturing gradient gel electrophoresis analysis given in Fig. 3. The similarity matrix is based on band position.

Throughout the experiment, the control and glucose-amended treatments shared more common bands with each other than they did with the naphthalene-amended treatment (Fig. 4). According to the UPGMA dendrogram, there were no significant similarities over time, but there were distinct patterns among treatments. The naphthalene-amended samples formed a separate cluster from the glucose-amended and the no-addition controls. There was no clear difference between the glucose-amended treatment and the no-addition control.

The dominant bands were excised and sequenced. Bands 1 to 10 were at least 95% similar to 16S rDNA gene sequences of bacteria belonging to either the Beta- (bands 1, 4–6, 9,10) or Gamma- (bands 2, 3 and 8) proteobacteria (Table 1). The sequences of bacteria belonging to the Gammaproteobacteria were most similar to two genera: Acinetobacter sp. (bands 2 and 3) and Pseudomonas sp. (band 8). Three of the bands recovered from the naphthalene-amended treatment were similar to known hydrocarbon degraders (Mueller et al., 1997; Razak et al., 1999): Pseudomonas lanceolota (band 10) and Acinetobacter sp. (bands 2 and 3). Three of the cultured bacteria capable of PAH degradation belonged to the Gammaproteobacteria. The 16S rRNA genes of the cultured bacteria were not similar to the uncultured bacteria sampled by DGGE (Table 1).

Table 1.   Identity of dominant bands detected by denaturing gradient gel electrophoresis in naphthalene (N), glucose (G), and no-addition control (C) incubations
Band numberIncubationClosest relativeSubdivision of
Accession number of
closest relative
% Similarity
  • *

    These bacteria are known to degrade aromatic hydrocarbons (Mueller et al., 1997; Razak et al., 1999).

  • The identities of three bacterial isolates from the naphthalene-amended treatment are also given.

1C,G,NVariovorax paradoxusBetaAF20946998
2NAcinetobacter sp.*GammaAY08350698
3NAcinetobacter sp.*GammaAY08350699
4C,G,NComamonas sp.BetaAF519533100
5NUnidentified bacteriumBetaAJ42216595
6C,G,NPseudomonas spinosaBetaAB021387100
8C,G,NPseudomonas aeruginosaGammaAF44803898
9C,G,NPseudomonas spinosaBetaAB02138798
10NPseudomonas lanceolota*BetaAB021390100
Isolate 1NArthrobacter sp.GammaAY247743100
Isolate 2NSphingomonas sp.GammaAY83837698
Isolate 3NSphingomonas sp.GammaAF361178100

Abundance of major phylogenetic groups

Four major phylogenetic groups, Alpha-, Beta-, and Gammaproteobacteria subgroups and the Cytophaga-Flavobacterium (CF) cluster, were investigated by FISH. Bacterial abundance increased in the control incubation (Fig. 5a), but the community composition was relatively stable throughout the experiment according to FISH (Fig. 5b). At time 0, Betaproteobacteria were more abundant than all other groups, but had decreased by day 3. The relative abundance of the other three bacterial groups also varied during the experiment, but all four groups were present at each time-point.

Figure 5.

 Total prokaryotic abundance as determined by 4′,6′-diamidino-2-phenylindole (DAPI) epifluorescence microscopy (a) and percent of DAPI-stained cells detected by fluorescent in situ hybridization in the control (b), glucose (c) and naphthalene (d) treatments. Eub is the general bacterial probe Eub338, Alpha, Beta, and Gamma are subdivisions of Proteobacteria, and “CF” is the Cytophaga-Flavobacteria cluster. Error bars indicate standard errors.

In contrast, the glucose-amended treatment was dominated by a few phylogenetic groups after the initial time-point (Fig. 5b). On day 3, Beta- and Gammaproteobacteria together comprised 40% of the DAPI-stained cells. By day 10, Betaproteobacteria were the most abundant group, while the abundance of Gammaproteobacteria had decreased to less than 5%. By day 25, Betaproteobacteria had decreased to undetectable levels. Both Alphaproteobacteria and the CF cluster were less than or equal to 10% of prokaryotic abundance throughout the experiment.

Even more than seen in the glucose-amended treatment, total bacterial abundance increased in the naphthalene-amended treatment (Fig. 5a), and the community was dominated by only a few phylogenetic groups (Fig. 5d). The Gammaproteobacteria increased by more than fivefold over the first 3 days of the experiment and accounted for the largest percentage of prokaryotes in the naphthalene-amended treatment (55%). At that time, Alphaproteobacteria and the CF group were not detectable. In contrast, Betaproteobacteria were twofold higher on day 10 than initially, and were the most numerous group to the end of the experiment, accounting for 24% of DAPI-stained cells. The Alphaproteobacteria were the only other group detectable by FISH on day 25.


This study found that bacterial production and abundance, naphthalene mineralization rates, and the ratio of naphthalene mineralization to bacterial production increased markedly in naphthalene-amended treatments. Bacterial production rates were probably high because naphthalene is a relatively labile carbon source. Naphthalene and other low-molecular-weight hydrocarbons are known to be good energy sources for some bacteria (Geiselbrecht et al., 1996). Although potentially toxic to bacteria at high concentrations (Langworthy et al., 2002), PAHs may support growth and even stimulate activity at a wide range of environmentally observed concentrations (Madsen et al., 1992). Unexpectedly, however, naphthalene had a more pronounced effect on bacterial activity than glucose. The in situ concentrations of glucose and naphthalene may explain this difference. Our study site in the Delaware estuary has relatively high PAH concentrations after having been exposed to PAHs over several years (Pohlman et al., 2002), a factor that has probably contributed to the capability of the microbial community not only to tolerate but also to be stimulated by naphthalene. In contrast, free glucose is undetectable in the Delaware estuary (Kirchman & Borch, 2003). It is conceivable that the glucose concentrations used here are less natural than the added naphthalene.

The ratio of naphthalene mineralization to bacterial production indicates that the bacterial community in the naphthalene-amended treatment had a greater capacity to degrade naphthalene than in the glucose-amended treatment or in the no-addition control. These enhanced naphthalene mineralization rates were associated with large shifts in bacterial community structure. In addition to culture-dependent studies, several previous culture-independent studies have also observed changes in microbial community structure after exposure to PAHs (Langworthy et al., 1998; Ringelberg et al., 2001; Hayes & Lovley, 2002; Del Panno et al., 2005). Using phospholipid fatty acid analysis (PLFA), Langworthy et al. (1998) found differences in microbial community composition of freshwater sediments with and without chronic exposure to PAHs. Ringelberg et al. (2001) also used PLFA analysis and found correlations between PAH degradation and the abundance of Rhodococcus spp. and actinomycetes. Based on DGGE and clone libraries of 16S rRNA genes, Hayes & Lovley (2002) suggested that Deltaproteobacteria, including a cultured representative, were involved in PAH degradation in anaerobic sediments.

Using DGGE we found that community structure changed as a result of the naphthalene amendments. Other DGGE studies have seen shifts in community composition associated with exposure of aquatic microbes to oil or PAHs (Macnaughton et al., 1999; Ogino et al., 2001), but these studies were unable to identify the affected taxa. Sequences of 16S rRNA genes from naphthalene-amended treatments in our experiment were similar to those from known hydrocarbon-degrading bacteria, although the short sequences available from DGGE bands do not allow a definitive identification. The phylotype represented by band 5, found only in the naphthalene-amended treatment, is not related to any known hydrocarbon-degrading bacteria; it is most similar to an uncultured bacterium found in a contaminated river (Brümmer et al., 2003). The enhancement of phylotypes in the naphthalene-amended treatment may indicate the presence of PAH degraders previously not identified by culture-dependent methods. Del Panno et al. (2005) recently used DGGE and other methods to examine the effect of petrochemical sludge, which included PAHs, on soil microbes. Their results are qualitatively similar to ours, although the bacteria affected by the contamination differ.

For a quantitative analysis of community composition, we used an independent, yet complementary method, FISH. The proportion of prokaryotic cells that we were able to detect with the eubacterial probe (Eub338) varied from 25% to 70%. The low percentages may be partly explained by methodological problems with FISH, including incomplete coverage of bacteria by the Eub338 probe (Daims et al., 1999). Another problem is detecting cells with low rRNA content due to low activity; the Eub338 counts and leucine incorporation rates were both lowest in the control and highest in the naphthalene-amended treatment. Archaea are not, of course, detected by bacterial probes; however, they are usually not abundant (<3%) in this environment (Kirchman et al., 2005). Although these problems may complicate the interpretation of the data, we still saw differences over time and between the naphthalene-amended treatment and the control.

Interestingly, entire phylogenetic groups became undetectable over 3 days in this experiment. The Cytophaga-Flavobacterium cluster, Gammaproteobacteria, and Alphaproteobacteria all varied in abundance and were undetectable during at least one sampling in the naphthalene-amended treatment. Experimental manipulations may reduce the diversity of microbial communities (Schafer et al., 2000), but the controls would be affected as well as the naphthalene-amended treatment. Exposure to hydrocarbon mixtures containing naphthalene has been shown to decrease bacterial diversity (Nyman, 1999), and the diversity of extreme or altered aquatic environments may be an order of magnitude lower than the diversity of pristine environments (Torsvik et al., 2002), although smaller reductions have also been observed (Andreoni et al., 2004). Even so, the loss in an entire phylogenetic group as observed with FISH in this study is surprising if these groups were diverse and consisted of thousands of phylotypes.

If the diversity of the bacterial community in this study was low, the reduction, or even loss, of an entire phylogenetic group seems more plausible. Relatively low diversity is supported by our DGGE data: DGGE indicated no more than eight dominant phylotypes per sample. Only two groups, Beta- and Gammaproteobacteria, were represented in the dominant bands we sequenced. The Alphaproteobacteria are not numerous or very diverse at our sample location (Kirchman et al., 2005). The Cytophaga-Flavobacterium cluster is generally abundant, but its high diversity (Kirchman et al., 2005) may complicate the detection of individual phylotypes by DGGE.

Regardless of the diversity within the bacterial groups examined here, it is clear that naphthalene had an effect on microbial communities in these experimental mesocosms. The reduction in the number of phylogenetic groups observed in naphthalene-amended treatments has potentially large ecological implications. These broad phylogenetic groups appear to differ in their degradation of organic material in aquatic ecosystems (Cottrell & Kirchman, 2000; Kirchman et al., 2004). Changes in the type or rate of organic matter use would greatly affect the fate of energy routed through the microbial loop. Naphthalene is a ubiquitous pollutant and may serve as an example of the potential effects of other organic pollutants. The development of a community enriched in naphthalene-degraders indicates the potential for bioremediation, but the lost diversity may have serious impacts on biogeochemical cycles and aquatic foodwebs (Girvan et al., 2005).


We thank Tom Boyd at the Naval Research Laboratory and the crew of the R/V Cape Henlopen. Support for shiptime was provided by the NRL Shiptime Program (to T. Boyd) and Office of Naval Research (N01403WX20068 to M.T.M.) This work was supported by a NOAA Sea Grant awarded to D.L.K.