Composition and dynamics of biostimulated indigenous oil-degrading microbial consortia from the Irish, North and Mediterranean Seas: a mesocosm study

Authors


Correspondence: Christoph Gertler, School of Biological Sciences, Environment Centre Wales, Bangor University, Bangor, Gwynedd LL57 2UW, UK. Tel.: +44 1248 383 056; fax: +44 1248 382 527; e-mail: c.gertler@bangor.ac.uk

Abstract

Diversity of indigenous microbial consortia and natural occurrence of obligate hydrocarbon-degrading bacteria (OHCB) are of central importance for efficient bioremediation techniques. To investigate the microbial population dynamics and composition of oil-degrading consortia, we have established a series of identical oil-degrading mesocosms at three different locations, Bangor (Menai Straits, Irish Sea), Helgoland (North Sea) and Messina (Messina Straits, Mediterranean Sea). Changes in microbial community composition in response to oil spiking, nutrient amendment and filtration were assessed by ARISA and DGGE fingerprinting and 16Sr RNA gene library analysis. Bacterial and protozoan cell numbers were quantified by fluorescence microscopy. Very similar microbial population sizes and dynamics, together with key oil-degrading microorganisms, for example, Alcanivorax borkumensis, were observed at all three sites; however, the composition of microbial communities was largely site specific and included variability in relative abundance of OHCB. Reduction in protozoan grazing had little effect on prokaryotic cell numbers but did lead to a decrease in the percentage of A. borkumensis 16S rRNA genes detected in clone libraries. These results underline the complexity of marine oil-degrading microbial communities and cast further doubt on the feasibility of bioaugmentation practices for use in a broad range of geographical locations.

Introduction

The recent Gulf of Mexico oil spill in 2010 has once again shown the urgent need for simple and robust bioremediation techniques that can be quickly implemented on a large scale. The unprecedented continuous flow of crude oil into the Gulf of Mexico presented a huge challenge to existing oil spill treatment methods, and current technologies were not able to cope with the size and nature of the oil spill. A few weeks after the closure of the well-head, only a small portion of the 575 million litres of oil was removed (Lubchenko et al., 2010), with 17% being recovered at the well-head, 3% skimmed off the water surface and 5% removed by in situ burning. The extensive use of highly toxic dispersants, such as Corexit® 9500, led to the further dispersal of 8% in the water body (Lubchenko et al., 2010). However, the release of the pressurized oil, together with the use of dispersants at the well-head, generated a 35 ×2 × 0.2 km oil plume at a depth of 1100–1300 m, where both water temperature and oxygen concentration are low (Camilli et al., 2010). A recent study investigated microbial populations within the oil plume, and although significant microbial growth was not detected, species of bacteria related to Oleispira spp. were found to be present (Hazen et al., 2010); however, ammonium, phosphate and oxygen levels were not significantly reduced. Chemical and natural dispersal procedures in the Gulf of Mexico emulsified about a quarter of the oil on the surface of the water (which increased the toxicity because of the higher bioavailability), while an estimated 26% of the oil remained untreated by the closure of the well-head 106 days after the spill began (Lubchenko et al., 2010).

Although bioremediation techniques have previously been successfully implemented, for example, following the M/V ‘Exxon Valdez’ oil spill (Lindstrom et al., 1991), there is still an urgent demand for the development and optimization of offshore bioremediation techniques that can play a central role in marine oil spill response contingency plans. In addition to the environmental advantages, bioremediation is economically highly competitive. It has been estimated that the clean-up costs of marine oil spills per litre of oil removed by bioremediation are 18.9 times lower than for manual clean-up, 9.2 times lower than skimming and mechanical removal, 4.6 times lower than dispersant application and 2.5 times lower than in situ burning (Etkin, 2000). Therefore, as part of an integrated and environmentally sustainable mitigation strategy, a more efficient bioremediation component could significantly reduce the costs of future oil clean-up operations.

One of the most important issues for a bioremediation technique is its universal applicability in different geographical locations. Sea water represents an almost ideal system to establish such a technique, as many of its abiotic factors, for example, salinity, pH, oxygen concentration and nutrient levels, show relatively little variation in contrast to other environments, like soil or fresh water. Because of the constant global exchange of sea water, taxonomically and functionally similar marine hydrocarbon-degrading microorganisms have evolved and spread worldwide. A distinct group of members of the Oceanospirillales have a high affinity towards oil hydrocarbon substrates. Species such as Alcanivorax borkumensis (Yakimov et al., 1998; Schneiker et al., 2006), Cycloclasticus pugettii (Dyksterhouse et al., 1995), Oleispira antarctica (Yakimov et al., 2003), Oleiphilus messinensis (Golyshin et al., 2002) and Thalassolituus oleivorans (Yakimov et al., 2004) constitute a distinct group of obligate hydrocarbon-degrading bacteria (OHCB) that outcompete most of the naturally occurring oligotrophic marine microorganisms following a sudden oil spill event (Hara et al., 2003; Yakimov et al., 2007). Some of these species have been linked to hydrocarbon-producing microalgae that may provide a natural niche for OHCB, which may explain the global distribution of these microorganisms (Green et al., 2004).

An obvious application would be the addition of a laboratory-grown culture of OHCB to an oil-impacted environment. However, the viability of these microorganisms after conservation procedures is significantly reduced, and for large-scale use, there is no adequate delivery system that can prevent microorganism dispersal. Furthermore, the microbial consortia required to degrade a specific blend of oil in a specific location is often too complex to be replaced by an artificial microbial culture mix (McKew et al., 2007); however, the enrichment of microorganisms in a complex system such as sea water often leads to an increase in predators, which could negate any possible benefits of bioaugmentation (Fu et al., 2009; Gertler et al., 2009b, 2010). An alternative to bioaugmentation is a highly efficient biostimulation strategy that delivers both marine oil-degrading microorganisms and nutrients at, or close to, the oil/water interface. Previous biostimulation studies have been carried out at a number of different scales, for example, oil-impacted sediments, micro- or mesocosm studies and bioremediation field trials (Lindstrom et al., 1991; Yakimov et al., 2005; Cappello et al., 2006; Alonso-Gutierrez et al., 2009; Gertler et al., 2009ab). Although the majority of these studies have reported large variations in microbial responses to oil pollution, they lack a common standardized method, which makes useful comparisons between them difficult.

The aim of this study was to determine the response of indigenous microbial consortia, from different geographical locations, to a simulated oil spill. We hypothesized that because of the global exchange of sea water and the relative stability of the seawater system, comparable oil-degrading microbial communities will be present across a broad range of geographical locations despite distinct site characteristics. Therefore, to test this hypothesis, we have used a standardized and simple mesocosm design, established with water from three different European marine sites: the Irish Sea, the North Sea and the Mediterranean. Identical crude oil was added to each mesocosm, together with a standardized slow-release fertilizer, and changes in microbial community composition were monitored by ARISA and DGGE fingerprinting and by 16S rRNA gene library analysis. As oil-degrading microbial consortia are susceptible to protozoan grazing, an additional set of mesocosms containing filtered water was established to investigate the effects of reduced grazing. All three sampling sites have a history of marine oil degradation studies (Cappello et al., 2006, 2007; Gertler et al., 2009ab, 2010) and show very different site characteristics of relevance to bioremediation techniques, for example, shipping numbers, temperature and nutrient availability.

Materials and methods

Experimental sites

The mesocosm experiments were conducted at the Biologische Anstalt at the Alfred Wegener Institute for Polar and Marine Research (AWI), Helgoland, Germany, using sea water from the Helgoland Eastern port site (54°11′N; 7°53′25″E) from 10 April 2009 to 10 May 2009; at the CNR-IAMC Institute for Coastal Marine Environment, Messina, Italy, using sea water from the Strait of Messina (38°11′53″N; 15°34′08″E) from 27 April 2009 to 21 July 2009; and at the School of Ocean Sciences, Bangor University, Menai Bridge, United Kingdom, using sea water from the Menai Straits (53°13′31″N; 4°09′33″E) from 27 July 2009 to 15 September 2009. To avoid any confusion of the Menai Straits and the Strait of Messina mesocosms, the Menai Straits site will henceforth be referred to as ‘Bangor’.

All three locations show very different meteorological and hydrological parameters. Hydrocarbon pollution in the Messina site is very high because of heavy shipping traffic in the Messina Strait but is lower at the Helgoland site because of less concentrated ferry traffic and to some extent recreational boating. The Menai Straits is affected by much less hydrocarbon pollution, as shipping activity is generally very low. Average sea surface water temperatures range from winter minimum temperatures of 3–5 °C and summer maximum temperatures of 17 °C for the Helgoland and Bangor sites and between 15 and 27 °C at the Messina site (El Hag & Fogg, 1986; Wiltshire & Manly, 2004; Azzaro et al., 2007).

Prior studies of all three sites have shown spring blooms of phytoplankton in April/May for Helgoland and Messina (Gillbrich, 1988; Azzaro et al., 2007) and in May/June for Bangor (Rodrigues & Williams, 2002). Concentrations of nutrients such as nitrate, phosphate and ammonium are subject to strong annual variations, which are only documented for the Helgoland site (Wiltshire et al., 2010). In general, the Helgoland site is significantly richer in nutrients than both the Bangor and the Messina sites (Rodrigues & Williams, 2002; Sitran et al., 2007; Wiltshire et al., 2010). Average nitrate concentrations range from 20 μM (Helgoland) (Raabe & Wiltshire, 2008) and 9 μM (Bangor) (Rodrigues & Williams, 2002) to 6 μM (Messina). Summer nitrate and ammonium concentrations are extremely low at all three sites. Phosphate concentrations are relatively similar in all locations with a concentration ranging from 0.8 μM at the Helgoland site (Raabe & Witshire, 2008), 1 μM at the Bangor site (Evans et al., 2003) to 0.35 μM at the Messina site (Sitran et al., 2007). Long-term monitoring data of 45 years are only available for the Helgoland site and show a substantial (1.13 °C) increase in average seawater temperature, together with decreases in the average concentration of ammonium, phosphate, silicate and nitrate. Ammonium and phosphate concentrations were reduced by approximately 80% and 50%, respectively, over this time (Wiltshire et al., 2010). With regard to the average nitrate and phosphate concentrations presented above, the N/P ratio at the Bangor (10 : 1.1) is generally higher than at the Messina (10 : 0.58) and Helgoland (10 : 0.5) sites and thus closer to the N/P ratio of 10 : 3 recommended for bioremediation (Atlas, 1981).

Mesocosm design and set-up

A standardized mesocosm experimental design was used, which was comprised of two treatments (filtered and unfiltered sea water) and three replicates of each mesocosm type. As mentioned earlier, to avoid any confusion of Menai Straits and Messina samples, the experiments conducted with water from the Menai Straits are referred to as ‘Bangor’ mesocosms. Mesocosms were contained in 50-L polypropylene barrels (Gaerner, Duisburg, Germany) filled with 30 L of filtered or unfiltered sea water. The filtered mesocosms (‘control mesocosms’) were filled with sea water that was passed through a nitrocellulose filter with 3-μm pore size (Sartorius, Göttingen, Germany) to remove or reduce the number of protozoa and thus reduce the effects of protozoan grazing. Unfiltered mesocosms (‘grazing mesocosms’) were filled with unfiltered sea water. Samples were taken from each mesocosm at 0, 3, 6, 9, 12, 15, 20, 25, 30, 40 and 50 days. During each sampling procedure, 5 L of water of each mesocosm was removed and replaced by 5 L of sterilized sea water. Collection of samples from the Helgoland mesocosms had to be stopped after 30 days, while for the Messina mesocosms, only two replicates could be performed for each type of mesocosm. To prevent the attachment of oil to the inner walls of the mesocosm and to decrease volatilization of the oil, an oil containment cylinder (OCC) manufactured at Bangor University was inserted into the barrel (Supporting Information, Fig. S1). The OCC consisted of a column of transparent polyvinylchloride at a height of 300 and 100 mm in diameter with a wall width of 0.5 mm. The cylinder was fitted with three openings of 100 mm in length and 20 mm in width for water circulation, which were cut into the cylinder at 120° angles 25 mm above the lower end of the cylinder. The upper end was closed by a transparent PVC lid of 100 mm diameter. To prevent the formation of air bubbles, a 2-mm ventilation hole was drilled 10 mm below the upper end of the top of the cylinder. Fifteen millilitres of Arabian light weathered crude oil, sterilized by filtration through a 0.2-μm syringe filter (Sartorius), was added to each OCC through the upper ventilation hole. For circulation in each mesocosm, an Eheim 1046 circulation pump (Eheim, Deizisau, Germany) was fitted outside the OCC into each mesocosm and connected to the OCC with a silicone tube. For biostimulation, 15 g of Miracle-Gro® NPK 18:9:11slow-release fertilizer (Scotts, Godalming, UK) was added into each mesocosm in this experiment. According to the manufacturer's description, this slow-release fertilizer contains 8% (1.2 g) nitrate, 10% (1.5 g) ammonium and 15.8% (2.37 g) phosphorus pentoxide. Therefore, 15 g of this fertilizer contains 102 mmol N and 16.7 mmol P. All microcosms were incubated at 16 °C in controlled temperature incubation rooms in Bangor, Helgoland and Messina.

Quantification of prokaryotic and protozoan populations by DAPI staining and fluorescence microscopy

Mesocosm water samples of 100 mL were taken at days 0 and 3 (and samples of 50 mL for all following sampling points), and 45 mL was fixed with 4.5 mL of 10% (w/v) paraformaldehyde and stored at 4 °C. For fluorescence microscopy, 10-mL aliquots of fixed samples were stained with 100 μL of a 10 mg mL−1 DAPI solution and filtered through black Isopore 0.2-μm polycarbonate filters (Millipore, Molsheim, France) using a bottle top filtration system (Sartorius). A 25 mm2 (5 × 5 mm) piece of each filter was removed using a sterile scalpel blade, embedded in Citiflour mounting solution (Citiflour Ltd., London, UK) and observed with an epifluorescence Zeiss Axioskop microscope (Zeiss, Jena, Germany). Using a net micrometer (12.5 × 12.5 mm, 5/5; 10, Zeiss), the cells in at least 200 microscopic fields were enumerated.

Extraction of nucleic acids

Microbial biomass was collected at every sampling point by filtration through a bottle top filtration system (Sartorius) with 47-mm cellulose acetate membrane filters (0.2-μm pore diameter) (Sartorius). Each filter was cut into pieces of approximately 2.5 × 2.5 mm and inserted into the lysis tubes of a Fast DNA kit (MP Biomedicals, Solon, OH). Lysis was performed with the FastPrep FP120 Cell disruptor (QBiogene, Carlsbad, CA), and DNA was extracted according to the manufacturer's instructions. DNA was resuspended in DNAse/RNAse-free ultrapure water (Fisher Scientific, Loughborough, UK) and stored at −20 °C.

PCR amplification of total nucleic acids for DNA fingerprinting analysis and DNA clone libraries

DNA extracts were diluted with nuclease-free ultrapure water to a concentration of 10–50 ng μL−1. One microlitre (10–50 ng of diluted extracts) was used as a template in PCR master mixes for both ARISA and DGGE and for the 16S rRNA clone library construction. The 20 μL of master mix for ARISA contained 2.5 mM MgCl2, 200 μM of each dNTP (Promega, Madison, WI), 10 nM of each primer 132f (5′-CCG GGT TTC CCC ATT CGG-3′) and 1522r (5′-TGC GGC TGG ATC CCC TCC TT-3′) (Ranjard et al., 2001), 2 μL of PCR Buffer B (Roboklon, Berlin, Germany), 1 U of Roboklon Taq DNA Polymerase (Roboklon) and 4 μL of a PCR-enhancing mixture (3 M betaine and 1% Tween 20). PCR amplification was performed as previously described (Gertler et al., 2009ab) using a DNA Engine Tetrad 2 Thermal Cycler (BioRad, Hercules, CA). The 50 μL master mix for DGGE contained 2.5 mM MgCl2, 200 μM of each dNTP, 10 nM of each primer 341f-GC (5′-CGC CCG CCG CGC CCC GCG CCC GGC CCG CCG CCC CCG CCC CCC TAC GGG AGG CAG CAG CCT ACG GGA GGC AGC AG-3′) and 907rm (5′-CCT ACG GGA GGC AGC AG-3′) (Muyzer et al., 1993), 5 μL of PCR Buffer B, 2 U of Roboklon Taq DNA Polymerase and 10 μL of the PCR-enhancing mixture. PCR amplification was performed as described previously (Sapp et al., 2007). The 50 μL master mix for DNA clone libraries contained 2.5 mM MgCl2, 200 μM of each dNTP, 10 nM of each primer 27f (5′-AGA GTT TGA TCC TGG CTC AG-3′) and 1492r (5′-GGT TAC CTT GTT ACG ACT T-3′), 5 μL of PCR Buffer B, 1 U of Roboklon Taq DNA Polymerase and 10 μL of the PCR-enhancing mixture. PCR amplification started with a denaturing step at 96 °C for 3 min and 30 cycles at 95 °C for 1 min, 53 °C for 1 min and 72 °C for 2 min followed by an extension step of 10 min at 72 °C and was performed in the DNA Engine Tetrad 2 Thermal Cycler (BioRad).

ARISA/DGGE fingerprinting analysis and multivariate statistics analysis

One ARISA profile for each individual DNA sample was produced using the Agilent DNA 1000 kit with the Agilent 2100 Bioanalyser (Agilent, Böblingen, Germany) according to the manufacturer's guidelines and the default program for the DNA 1000 chip. Data were saved as gel images and as band-matching tables containing band intensity values for multivariate statistical analysis. Normalization of band patterns was conducted automatically using the internal length standard of the Agilent DNA 1000 Kit, and a total of 56 band classes were assigned according to the length of the PCR products in each sample. For DGGE fingerprints, gel pictures were normalized according to the Roboklon 1 kb plus Standard (Roboklon) and analysed using Quantity One Software (BioRad). A total of 24 band classes were assigned to DGGE bands. Densitometric values for each ARISA or DGGE band were included in the analysis, resulting in a band-matching table. Square root transformation was used to calculate Bray–Curtis similarity of the ARISA and DGGE fingerprints (Clarke & Warwick, 2001) for each fingerprint and the whole data set. Ordination of similarity matrices was conducted by principal coordinate analysis (Clarke & Warwick, 2001) using Primer 6 software (Clarke & Gorley, 2001). Permutational multivariate analysis of variation (permanova) was performed using the permanova+ add-on for primer 6. permanovas were conducted on the basis of the Bray–Curtis distance measure. P-values were obtained using 9999 permutations of residuals under a reduced model.

DGGE fingerprinting analysis, band excision and re-amplification

For DGGE fingerprinting analysis, one replicate from each sampling site and from each treatment was selected based on uniformity of each mesocosm pair observed in the ARISA fingerprints. The mesocosms chosen for analysis were Bangor Control 1, Bangor Grazing 3, Helgoland Control 3, Helgoland Grazing 1, Messina Control 2 and Messina Grazing 2. The same genomic DNA used in ARISA fingerprinting was used in the subsequent DGGE-PCR. Denaturing gradient gel electrophoresis was conducted following the protocol of Muyzer et al. (1993) using an Ingeny PhorU Electrophoresis System (Ingeny, Goes, Netherlands). DGGE gels contained 6% (w/v) polyacrylamide denaturing gradient gels with linear gradients from 40% to 60% of denaturing agents [where 100% denaturant is 7 M urea and 40% (v/v) formamide]. According to the signal intensity of each PCR product on 1% agarose gels, 10–50 ng of each PCR product was loaded onto the DGGE gel and run in 1× TAE buffer at 100 V, 200 mA at 60 °C for 16 h. Gels were stained with a 1× SYBR-Gold (Invitrogen, Carlsbad) solution in 1× TAE buffer and visualized with a BioRad XRS gel documentation system (BioRad). Band excision was performed on an illumination table with disposable scalpel blades. Resulting acrylamide blocks were soaked in 20 μL of nuclease-free ultrapure water in a 1.5-mL Eppendorf tube and incubated in an Eppendorf Thermomixer (Eppendorf, Hamburg, Germany) at 37 °C for 12 h. PCRs were performed as described previously using 1 μL of each suspension as template. The resulting PCR products were tested for purity by further DGGE analysis.

DNA clone libraries

On the basis of the results of both ARISA and DGGE fingerprinting, we performed an analysis of SSU rRNA clone libraries established from climax communities in the six mesocosm parallels previously selected for DGGE analysis. Clone libraries were constructed from DNA samples taken from these mesocosm parallels at day 25, which showed stable oil-degrading consortia in all six cases in both ARISA and DGGE fingerprints. As before, the same genomic DNA previously used in both fingerprinting techniques was used in the subsequent PCR described earlier. PCR products were purified using the MinElute PCR purification kit (Qiagen, Hilden, Germany), and 20 ng of each was ligated into PCR 2.1 vector using the TOPO TA cloning kit. Ligation was performed at room temperature for 30 min according to the manufacturer's manual. DNA was precipitated with n-butanol and the resulting pellet air-dried and resuspended in 5 μL of sterile molecular grade water. Ligation products were added to 25 μL of Electromax DH10B competent cells (Invitrogen), and electroporation was performed with an ElectroPulser electroporator (BioRad) at 1.8 kV for 4 ms. Cells were recovered in 250 μL of LB medium prewarmed at 37 °C for 1 h, and after that, 25 μL of each cloning reaction was plated onto LB agar plates containing 25 mg L−1 kanamycin and 25 mg L−1 X-Gal for subsequent blue/white colony selection. White clones were picked and cultured overnight at 37 °C in deep-well plates (Nunc, Denmark) containing LB medium and 25 mg L−1 kanamycin. Cloned inserts were analysed by PCR of culture material from single well with the vector-specific M13 forward and reverse primers (Invitrogen). The resulting PCR products were purified using the MinElute 96 PCR Purification Kit (Qiagen) according to the manufacturer's instructions. PCR product sequencing was carried out with the BigDye Terminator v1.1 Kit (Applied Biosystems, Foster City, USA) as described in the manufacturer's manual. The sequencing reaction was performed on an Eppendorf Mastercycler (Eppendorf, Hamburg, Germany) using the primers 530f (5′-GCT CTA GAG CTG ACT GAC TGA GTG CCA GCM GCC GCG G-3′) and 1492r (5′-GGT TAC CTT GTT ACG ACT T-3′), starting with an initial denaturing step of 96 °C for 3 min and 35 cycles of 96 °C for 1 min and 50 °C for 30 s and 60 °C for 1 min. The reaction product was cleaned up using the DyeEx 96 Kit (Qiagen) and the reaction products dried in a Savant vacuum concentrator (GMI, Ramsey, USA). Pellets were resuspended in 20 μL formamide and sequenced using a Perkin Elmer capillary sequencer (Applied Biosystems).

Phylogenetic analysis of DNA sequences from DGGE gels and DNA clone libraries

Sequences from both DNA strands were assembled using bioedit (version 2.5) (Hall, 1999). All sequences were checked for chimeras using the Pintail 1.1 program (Ashelford et al., 2005). Construction of phylogenetic trees and bootstrapping were performed using bioedit, clustalw and mega4.0 as previously described (Gertler et al., 2010). Nonredundant DNA sequences obtained from this study were deposited in the GeneBank/DDBJ/EMBL under accession numbers HE572626HE572699 (DNA sequences obtained from DNA clone libraries) and HE572700HE572745 (DNA sequences obtained from excised DGGE bands).

Results

Microbial and protozoan population dynamics

Both bacterial and protozoan populations showed comparable trends for all three experimental sites. Bacterial numbers ranged from 104 to 105 cells mL−1 at the beginning of the experiment, before quickly increasing to a maximum of between 2 × 106 and 107 cells mL−1 (Fig. 1) within the first 10 days of the experiment. However, bacterial populations remained fairly stable for the rest of the experiment. In the Bangor Grazing and Helgoland Control mesocosms, a significant decrease in prokaryotic cell counts was observed after 6 days following the initial microbial growth phase. Protozoan cell counts were monitored for both flagellates and ciliates because of their possible dual predation lifestyles observed in a prior study (Gertler et al., 2010). However, differences in cell numbers between the filtered and nonfiltered mesocosms at each individual experimental site were not significant. Both flagellate and ciliate cell counts showed large fluctuations in population densities during the course of the experiment. The Helgoland filtered mesocosms contained very few ciliates but a relatively constant flagellate population of approximately 103 cells mL−1. The corresponding grazing mesocosms showed ciliate populations of up to 100 cells mL−1, which was reciprocally proportional to the flagellate population of 103–5 × 103 cells mL−1. Similar trends could be observed for the Bangor mesocosms, with significantly elevated ciliate population sizes of up to 5 × 103 cells mL−1. In contrast, the Messina mesocosm had both flagellate and ciliate populations in an order of magnitude higher than at the other sites.

Figure 1.

Prokaryotic (bacterial and archaeal) (filled circles), flagellate (open triangles) and ciliate (filled squares) cell numbers in mesocosms containing filtered (Control) or unfiltered (Grazing) water. Data points represent the mean of three replicate mesocosms ± the standard error of the mean. For clarity, lower error bars of some data points are not shown.

Dynamic changes in bacterial community compositions

Microbial community dynamics were assessed by both ARISA and DGGE. Principal coordinate analysis of the ARISA fingerprints showed a convergence of fingerprints for all three locations between day 15 and 40 in a very small area of the Principle Coordinate Analysis (PCO) plot (Fig. 2a), indicating an extremely high similarity between all locations and treatments during this period, whereas samples obtained before or after this period showed a trend for location specificity rather than treatment specificity. This effect is more apparent in the individual fingerprints shown in Fig. S2. In contrast, the PCO plot of the DGGE fingerprints shows much less sample clustering (Fig. 2b), with a trend towards a pattern of location specificity. However, a trend towards sample clustering can only be seen for the Bangor samples (both treatments). All of the ARISA fingerprinting profiles showed a significant similarity in oil-degrading community structure at all sites and between filtered and unfiltered mesocosms. In all six cases, a mature oil-degrading community was established within the first 2 weeks. The time needed for the establishment of this microbial consortium was 6–12 days at the Helgoland site, 6–9 days at the Bangor site and less than 3 days in Messina (Fig. S2). These microbial communities remained stable until the termination of the experiment for Helgoland and Bangor, although a return to a more diverse community was observed between day 30 and 40 at the Messina site. Fingerprinting profiles contained very similar, or identical, band patterns in all of the mesocosms between day 15 and 30 (Fig. 3c). However, subtle differences between the fingerprinting patterns of the individual mesocosms were detected, suggesting slight differences in microbial compositions between replicates. There were no significant differences in the fingerprinting patterns between unfiltered and filtered mesocosms. permanova of both ARISA and DGGE fingerprints indicated significant differences in sample sets from each site (Tables S1 and S4). However, only the ARISA and DGGE data sets for the Helgoland samples showed significant differences with regard to filtration treatment (Tables S2 and S5). Pairwise tests of the complete ARISA data set for this experiment revealed that significant differences between locations could be observed particularly on day 3 and on days 40 and 50 (Table S3).

Figure 2.

Statistical analysis of DNA fingerprints: (a) PCO plot of ARISA fingerprints. Fingerprint patterns of both treatments at all three sites were applied in the analysis; (b) PCO plot of DGGE fingerprints. Fingerprint patterns from one mesocosm for each treatment (Bangor Control 1, Bangor Grazing 3, Helgoland Control 3, Helgoland Grazing 1, Messina Control 2 and Messina Grazing 2). Samples from Bangor are indicated by diamonds, samples from Helgoland represented by triangles and samples from Messina are shown by squares.

Figure 3.

DGGE profiles of microbial community changes in crude oil-degrading microbial consortia. Template DNA was extracted from the water body of the mesocosms at time points indicated on the bottom of the gel. The upper gel shows the microbial communities from both ‘ungrazed’ control mesocosms containing water from Bangor or Helgoland, and the middle gel shows the microbial communities for both ‘grazing’ mesocosms for Bangor and Helgoland. The lower gel shows both ‘ungrazed’ control' and ‘grazing’ Messina mesocosms. The mesocosm parallels chosen for the DGGE fingerprints were Bangor Control 1, Bangor Grazing 3, Helgoland Control 3, Helgoland Grazing 1, Messina Control 2 and Messina Grazing 2. Unmarked lanes contain a DNA size marker (Perfect 1kB plus® Marker (Roboklon) for upper and middle gel, and 1 kB DNA ladder® marker (Promega) for lower gel.

Table 1 shows the results of DGGE band sequencing with the majority of the excised band sequences belonging to members of the Alphaproteobacteria, Gammaproteobacteria and the Flavobacteria. Four prominent bands occurred in all six mesocosms (Bands C3, 6, 7, 8; G1, 2, 3, 6; M 4, 5, 6, 7) and showed high similarity to the 16S rRNA gene of A. borkumensis SK2. A 16S rRNA gene sequence with a 94% similarity to O. antarctica was detected in DGGE band G11, which exclusively occurred between day 12 and 30 in the ‘grazing’ mesocosm from the Bangor site. Further, 16S rRNA gene sequences similar to those of organisms with capabilities of degrading hydrocarbons were detected in bands M9 and M12. These bands occurred exclusively in the fingerprinting profiles of the Messina mesocosms between days 30 and 50. Two bands connected to members of the Alphaproteobacteria were abundant within all sites and treatments during most of the experiment. Band C5 contained DNA sequences with a 98% similarity to the 16S rRNA gene of a Phaeobacter sp., which was present in both types of mesocosms from the Helgoland site between days 9 and 30, and between days 3 and 20 of both types of Bangor mesocosms. However, it was not seen in the Messina mesocosms, despite band M11 being similarly placed within the banding pattern. Band M11 was identical to the 16S rRNA gene of a Roseobacter sp., which could be detected at day 9–30 in both Messina mesocosms.

Table 1. 16S rRNA gene sequences from DGGE bands
DGGE bandClosest hitAccession numberPhylogenySequence identity (%)
  1. The Prefix code B, H and M represent the experimental sites Bangor, Helgoland and Messina. The prefix code C and G represent the experimental treatment: ‘control’ – C (filtered water) or ‘grazing’ – G (unfiltered water). Accession number of the closest relatives of the DNA sequences determined with the megablast subroutine of blast tool.

C1Uncultured Rhodobacteraceae bacterium FM958455 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 96
C2Carbazole-degrading bacterium CAR-SF AB086227 Proteobacteria; Gammaproteobacteria; Oceanospirillales 98
C3 Alcanivorax borkumensis SK2 NR_029340 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Alcanivoracaceae 98
C4 Sulfitobacter litoralis DQ097527 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 99
C5 Phaeobacter sp. GQ906799 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 82
C6 Alcanivorax borkumensis SK2 NR_029340 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Alcanivoracaceae 98
C7 Alcanivorax borkumensis SK2 NR_029340 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Alcanivoracaceae 99
C8 Alcanivorax borkumensis SK2 NR_029340 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Alcanivoracaceae 99
C9 Thalassobius mediterraneus AB47094 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 87
C10Uncultured actinobacterium GQ349556 Actinobacteria 89
C11Uncultured Polaribacter sp. DQ473564 Bacteroidetes; Flavobacteria; Flavobacteriales; Flavobacteriaceae 84
C12Uncultured Bacteroidetes/Chlorobium group bacterium FJ615164 Bacteroidetes 80
C13 Hoeflea sp. MH129 EU052704 Proteobacteria; Alphaproteobacteria; Rhizobiales; Phyllobacteriaceae 97
C21Uncultured Thiocapsa sp. AJ632077 Proteobacteria; Gammaproteobacteria; Chromatiales; Chromatiaceae 90
G1 Alcanivorax borkumensis SK2 NR_029340 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Alcanivoracaceae 98
G2 Alcanivorax borkumensis SK2 NR_029340 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Alcanivoracaceae 98
G3 Alcanivorax borkumensis SK2 NR_029340 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Alcanivoracaceae 97
G4Uncultured Rhodobacteraceae bacterium FM958455 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 99
G5 Sulfitobacter litoralis DQ097527 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 100
G6 Alcanivorax borkumensis SK2 NR_029340 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Alcanivoracaceae 99
G7 Roseobacter sp. PIC-68 AJ534238 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 99
G8 Phaeobacter arcticus DQ514304 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 98
G9 Hoeflea siderophila EU670237 Proteobacteria; Alphaproteobacteria; Rhizobiales; Phyllobacteriaceae 95
G10 Hoeflea sp. MH129 EU052704 Proteobacteria; Alphaproteobacteria; Rhizobiales; Phyllobacteriaceae 98
G11 Oleispira antarctica NR_025522 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Oleispira 94
G12Uncultured gamma proteobacterium GQ348661 Proteobacteria; Gammaproteobacteria 99
G13 Roseobacter sp. AF098493 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 99
G14Uncultured Thiocapsa sp. AJ632077 Proteobacteria; Gammaproteobacteria; Chromatiales; Chromatiaceae 84
G15 Roseobacter pelophilus AJ968651 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 86
G16 Rhodobacteraceae bacterium GQ468664 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 86
G17Uncultured gamma proteobacterium GQ348736 Proteobacteria; Gammaproteobacteria 96
G22Carbazole-degrading bacterium CAR-SF AB086227 Proteobacteria; Gammaproteobacteria; Oceanospirillales 96
M1 Roseobacter sp. EU195946 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 99
M2 Roseobacter sp. AJ534238 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 98
M3 Roseobacter sp. EU195946 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 98
M4 Alcanivorax borkumensis SK2 NR_029340 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Alcanivoracaceae 100
M5 Alcanivorax borkumensis SK2 NR_029340 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Alcanivoracaceae 99
M6 Alcanivorax borkumensis SK2 NR_029340 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Alcanivoracaceae 99
M7Uncultured Chromatiales bacterium EU375149 Proteobacteria; Gammaproteobacteria; Chromatiales; Ectothiorhodospiraceae 99
M8 Alcanivorax borkumensis SK2 NR_029340 Proteobacteria; Gammaproteobacteria; Oceanospirillales; Alcanivoracaceae 98
M9 Muricauda lutimaris EU156065 Bacteroidetes; Flavobacteria; Flavobacteriales; Flavobacteriaceae 96
M10 Sneathiella glossodoripedis AB289439 Proteobacteria; Alphaproteobacteria; Sneathiellales; Sneathiellaceae 89
M11 Roseobacter sp. EU195946 Proteobacteria; Alphaproteobacteria; Rhodobacterales; Rhodobacteraceae 100
M12 Parvibaculum lavamentivorans CP000774 Proteobacteria; Alphaproteobacteria; Rhizobiales; Phyllobacteriaceae 99
M13 Tistrella sp. EU375009 Proteobacteria; Alphaproteobacteria; Rhodospirillales; Rhodospirillaceae 100

Composition of the oil-degrading community

DNA clone libraries were constructed to analyse the composition of the oil-degrading consortia and the percentage of individual species within the microbial communities. Clone libraries contained 122 and 116 clones for the Bangor (Irish Sea) ‘control’ and ‘grazing’ mesocosm, respectively, 123 and 101 clones for the corresponding Helgoland mesocosms, and 109 and 114 clones for the Messina mesocosms. In accordance with the DNA fingerprinting analysis, DNA clone libraries confirmed a high proportion of A. borkumensis-like sequences. The 60 randomly selected DNA clone sequences were compared with a database of 16S rRNA gene sequences from Alcanivorax species and revealed that the organisms detected in mesocosms from all three sites were identical to the type strain A. borkumensis SK2 (Fig. S3). In all cases, sequences from A. borkumensis-related organisms were present in the clone libraries from each site and treatment, with clone percentage ranging from 32.5% to 93.4% of the total clone numbers (Fig. 4). At the class level, the largest fraction of clones in all six cases belonged to the Gammaproteobacteria followed by Alphaproteobacteria with relatively low percentages of Flavobacteria, and several others. There was a reduction in Gammaproteobacteria-derived clone numbers in the ‘grazing’ mesocosms compared to the control mesocosms, together with a concurrent increase in the percentage of clones from Alphaproteobacteria, mostly derived from members of the family Roseobacteriaceae. Three highly similar DNA sequences were found in the mesocosms at all three locations, which corresponded to the species Aborkumensis, Cycloclasticus spirillensus and Phaeobacter arcticus. Sequences similar to those of P. arcticus were also very similar to the 16S rRNA gene sequence of the Roseobacter spp. detected in DGGE band M11 referred to as Roseobacter sp. 73 in an earlier study on oil-impacted sediments in northern Spain (Alonso-Gutierrez et al., 2009). The construction of a maximum parsimony phylogenetic tree linked this organism with two Roseobacter spp. and several Phaeobacter spp. (Fig. S3). In common to all three sites, the percentage of clones derived from A. borkumensis was reduced, while the clones derived from the Rhodobacteriaceae and Cytophaga-Flavobacterium-Group increased in the nongrazed compared to the grazed mesocosms (Fig. 4). In the Bangor mesocosms, the Phaeobacter-like Alphaproteobacteria were replaced by other representatives of the genus Roseobacter. Similarly, the ratio of detected species per clone library was reduced in all grazing mesocosms in comparison with their corresponding control mesocosms, indicating a higher microbial diversity in the absence of grazers. The percentage of A. borkumensis-like sequences within the oil-degrading consortia was highest in the control mesocosms at the Bangor site, with decreasing levels at the Helgoland and the Messina site. However, the percentage of above sequences was similar in the grazing mesocosms from the Bangor and Helgoland sites, but was about 50% lower at the Messina site. The clone library from the Bangor grazing mesocosm contained sequences derived from four different OHCB, A. borkumensis, C. spirillensus, Thalassolituus oleivorans and O. antarctica, and although C. spirillensus was observed at all three sites, the Bangor grazing mesocosms had the greatest diversity of OHCB than the other two sites (Fig. 4). This confirmed the data obtained from the DGGE fingerprinting analysis that detected O. antarctica in DGGE band G11 that exclusively occurred in the Bangor grazing mesocosm. Shannon and Simpson indices for all clone libraries were calculated with PAST software (Hammer et al., 2001) and showed significantly increased values for the grazing mesocosms compared to the nongrazed mesocosms. The Shannon indices were 0.335/1.2 for the Bangor mesocosm (‘control’/'grazing'), 1.03/1.39 for the Helgoland and 1.13/1.65 for the Messina site (Table S7). Simpson indices were 0.138/0.52 for the Bangor mesocosm (‘control’/'grazing'), 0.364/0.527 for the Helgoland and 0.494/0.733 for the Messina site (Table S7). Therefore, Shannon indices were lower in filtered mesocosms at each respective site, indicating a lower evenness of the filtered treatments. Conversely, Simpson indices for filtered mesocosms of each site were significantly higher than in unfiltered treatments.

Figure 4.

Percentage of OTUs obtained in the clone libraries. The Prefix code B, H and M represent the experimental sites Bangor, Helgoland and Messina. The prefix code C and G represent the experimental treatment: ‘control’ – C (filtered water) or ‘grazing’ – G (unfiltered water). Name attributed to the OTUs are those of closest related and validly published species or genus according to blast search.

Discussion

Microbial community population dynamics

The microbial population densities in the mesocosm from all three locations were very similar. The initial increase in cell number is common in marine oil degradation studies because of the effect of a sudden increase in carbon levels (Kasai et al., 2002a; Cappello et al., 2006; Gertler et al., 2009b), which leads to decreases in the oligotrophic microbial population in favour of the faster-growing marine OHCB. In natural environments, these microorganisms can reach extraordinarily high concentrations that are more characteristic of laboratory cultures (Kasai et al., 2002b). Although there was a high consistency in the data from all three locations, and in both ungrazed and grazed mesocosm, there were some important differences to previous studies that have been carried out at the same locations. A mesocosm study performed earlier at the Helgoland site at identical experimental temperatures (Gertler et al., 2009b) revealed microbial population densities in an order of magnitude higher than that observed in the present study. The reason for this could be the different amount and type of oil, together with a higher load of nitrogen/phosphorus fertilizer in the earlier study. Similar microcosm and mesocosm studies from the Messina site have also had significantly higher abundances of microorganisms, ranging from 108 to 1010 cells mL−1 (Cappello et al., 2006). These studies, together with data from field-scale experiments (Kasai et al., 2002b), highlight the difficulty in comparing results obtained from different experimental designs. However, analysis of microbial communities with both ARISA and DGGE fingerprinting techniques has demonstrated high levels of similarity between the three geographic sites between days 6 and 40, which is in part attributed to the abundance of Alcanivorax spp. (Fig. 5). No significant differences in the fingerprint patterns in the second week of the experiment could be detected (Fig. 2a) in comparison with the initial and final stages, which supports earlier observations from a variety of microcosm and mesocosm studies at both the Helgoland and Messina sites despite the use of different types of oil being applied (Cappello et al., 2006; Gertler et al., 2009ab). A comparison of 16S rRNA gene sequences (Fig. S3) of all Alcanivorax-related OTUs detected in both clone libraries and DGGE bands was identical or had an identity above 98% with the type strain A. borkumensis SK2. The overwhelming prevalence of microorganisms from the genus Alcanivorax in oil-degrading microbial consortia in temperate zones is well established.

Figure 5.

Rooted phylogenetic tree of 16S rRNA gene sequences from DGGE bands and clone libraries. The Prefix code B (diamonds), H (triangles) and M (squares) represent the experimental sites Bangor, Helgoland and Messina. The prefix code C and G represent the experimental treatment: ‘control’ – C (filtered water) or ‘grazing’ – G (unfiltered water) Prefix codes were amended by the genera of the closest relatives of the OTUs according to a blast search and are printed in bold letters. Sequences were clustered by Neighbour joining of maximum likelihood values showing affiliation of 917 bp partial prokaryotic according to the recommendation of the modeltest software (Posada & Crandall, 1998). 16S rRNA gene sequences to closest related sequences from either cultivated or cloned members of different lineages. The evolutionary distances were computed using the neighbor-joining and Jukes-Cantor methods in mega4 (Tamura et al., 2007) and are in the units of the number of base substitutions per site on the scale bar given for reference.

This microorganism is a very effective primary colonizer of oil/water interfaces in temperate zones because of a number of adaptations: multiple alkane hydroxylases, strong mineral nutrient scavenging and biosurfactant production and biofilm formation capabilities (Abraham et al., 1998; Schneiker et al., 2006). However, one of the limitations of this microorganism may be its susceptibility to low temperatures as it rarely occurs in oil degradation studies of polar zones (Yakimov et al., 2007) where psychrophilic OHCB such as O. antarctica seem to play a pivotal role (Yakimov et al., 2003; Brakstad & Lodeng, 2005). Sequencing of DGGE bands and SSU rRNA gene clone libraries showed abundance of this microorganism exclusively at the experimental site in Bangor, which suggests that both Alcanivorax spp. and Oleispira spp. may coexist and that their geographic distribution overlaps.

Interestingly, the DNA clone library data for the Bangor site revealed the presence of four different OHCB genera, Alcanivorax, Oleispira, Thalassolituus and Cycloclasticus, the latter being a well-known, and globally distributed, aromatic hydrocarbon-degrading specialist (Kasai et al., 2002a; Yakimov et al., 2007). Similar microorganisms capable of degrading a complex crude oil hydrocarbon mixture have previously been described in microcosm studies using water from the estuaries of the rivers Thames and Seine enriched with individual oil constituents (McKew et al., 2007; Niepceron et al., 2010), implying that the diversity of OHCB was mainly owing to the specialization of OHCB to different chemical components of the oil. Although we had no negative control mesocosms in this study (without oil spiking or nutrient amendment), the higher abundance of OHCB in the mesocosms was very likely due to the elevated levels of hydrocarbons as well as the nutrients provided by the slow-release fertilizer. In a previous study, we have shown that a negative control microcosm contains no elevated cell numbers or nucleic acid concentrations (Gertler et al., 2009a). In the absence of fertilizers, microbial communities remain very diverse, and the provision of nitrogen and phosphorus strongly selects for A. borkumensis. Our results also indicate that location can have an influence on the community composition, that is, only three microbial genera occurred at all three sites: Alcanivorax, Cycloclasticus and an alpha-proteobacterium related to genera Roseobacter/Phaeobacter previously detected in marine sediments polluted by oil in the aftermath of the M/V ‘Prestige’ oil spill (Alonso-Gutierrez et al., 2009). Similar to the 16S rRNA gene sequences of Alcanivorax-related OTUs (Fig. S3), sequences corresponding to both Cycloclasticus sp. and the alpha-proteobacterium mentioned above were detected in clone libraries as well as in DGGE bands and were identical or had a similarity above 99% (Fig. S4).

Alphaproteobacteria are commonly found in association with OHCB-dominated consortia, and while a number of them, for example, Thalassospira tepidiphila and Tropicibacter oleivorans (Kodama et al., 2008; Harwati et al., 2009), are capable of degrading aromatic hydrocarbon, the role of the Roseobacter spp. detected in this study remains unclear. Previous mesocosm studies have detected similar Alphaproteobacteria species in biostimulated oil-spiked mesocosms, but not in mesocosms containing no added nitrogen and phosphorus (Roling et al., 2002). Comparison of DGGE fingerprinting profiles of all treatments and sites showed that two ubiquitous species of Alphaproteobacteria occurred at the early stages of the experiment and remained detectable until the end of the experiment. The Messina mesocosms yielded high numbers of clones from Alphaproteobacteria (related to P. arcticus) from day 3 until the end of the experiment, whereas the Helgoland mesocosms contained species more closely related to a Roseobacter spp., which had originally been isolated at the same site by enrichment in a tryptone-yeast extract broth (Allgeier et al., 2003). These studies strongly suggest that these members of the Rhodobacteraceae are copiotrophic organisms benefitting from the nutrient enrichment in the oil-spiked and N/P amended mesocosms.

Composition of oil-degrading microbial consortia

While a common set of microorganisms essential for oil degradation were found at all sites, the majority of the microbial community was site specific; however, the role of these site-specific microorganisms during oil degradation remains unclear. A comparison of the ARISA and DGGE profiles indicated that at the chosen time points, microbial communities were relatively constant with few differences in the banding patterns. DNA clone libraries provided a further opportunity to analyse the microbial communities in all six treatments, after an identical incubation time, and generally supported the conclusions drawn from the DGGE fingerprinting that all microbial communities were dominated by Gammaproteobacteria and by A. borkumensis in particular. However, in contrast to the fingerprinting results, clone libraries showed a significant reduction in both evenness and diversity in all filtered treatments compared to unfiltered treatments, which was probably due to the dominance of the microbial community by A. borkumensis. The mesocosms from the Messina site showed a very high abundance of C. spirillensus, which may result from the higher pollution levels at this location. DNA from the Bangor mesocosms contained sequences with high similarity to four different OHCB, including two members of the Alphaproteobacteria (P. arcticus, Roseobacter spp.). DGGE fingerprinting demonstrated that the Phaeobacter-like microorganism was highly abundant in the Messina samples, less abundant in the Helgoland samples and nearly absent in the Bangor site, whereas numbers of the Roseobacter-like bacterium were elevated in the mesocosms from this site. Several members of the Flavobacteria were detected in all clone libraries, some species of which are known to degrade aromatic and polyaromatic hydrocarbons. Therefore, it is likely that they are capable of degrading intermediates following primary oil degradation by OHCB, for example, species of Flavobacteria with possible links to polyaromatic hydrocarbon (PAH) degradation have been reported in two prior studies performed at Helgoland, which used a heavy fuel oil rich in aromatic compounds (Gertler et al., 2009ab); furthermore, Flavobacteria of the genus Tenacibaculum and Olleya have been linked with predation (Banning et al., 2010). Bacterial predation in a comparable experimental system by Bacteriovorax spp. was detected during biostimulation experiments at the Helgoland site following the growth of OHCB (Gertler et al., 2009b).

Two ubiquitous OHCB were identified as Alcanivorax spp. and Cyclocasticus spp. representing the genera involved in primary degradation of aliphatic and aromatic hydrocarbons, respectively. The third organism was identified as a Phaeobacter-like member of the Rhodobacteraceae, which has previously been detected in oil-impacted sediments. Surprisingly, the 16S rRNA gene sequences from those organisms were identical regardless of geographical location they came from.

Effects of reduced protozoan grazing

Bioremediation relies on the in situ enrichment of microorganisms beyond the size of a naturally occurring population, although further increases in population size will be accompanied by a corresponding increase in predation. This predation can involve bacteriophages (Šimek et al., 2007; Zhang et al., 2007; Fu et al., 2009), predatory bacteria (Gertler et al., 2009b; Banning et al., 2010) or protozoa (Kota et al., 1999; Dalby et al., 2008; Gertler et al., 2010); however, little is known about the ecology of predation of OHCB microbial communities, particularly on a more global scale. Predation is capable of slowing down or even impeding bioremediation attempts (Fu et al., 2009), and although previous studies have shown the occurrence of protozoan predation at the Bangor and Helgoland sites (Andrews & Floodgate, 1975; Gertler et al., 2010), the effect of filtration in the current study had no effect on the occurrence of marine protozoa. Importantly, filtration through a 3-μm-pore-size filter significantly reduced the numbers of larger protozoa such as ciliates, but was unable to remove all protozoans, for example, heterotrophic nanoflagellates (HNF) that retained their initial numbers. Despite the removal of large protozoa, the microbial populations in both filtered and unfiltered mesocosms did not show significant differences at any of the three locations. Flagellate and ciliate cell counts for both filtered and unfiltered mesocosms at the Helgoland and Bangor sites did show a trend towards a cyclic curve progression similar to a predator–prey relationship. The interactions between flagellates and ciliates in marine systems are very complex, and previous mesocosm experiments at the Helgoland site have indicated that flagellate and ciliate abundance is correlated with prey size (Gertler et al., 2010). This competitive exclusion mechanism could be due to the predator cell size (Matz & Kjelleberg, 2005), predator feeding style (Tso & Taghon, 1999), predator energy conversion efficiency (Dini & Nyberg, 1999) or any combination of these factors.

Antibiosis of flagellates and ciliates limits the predation of HNF, which are able to grow faster than ciliates, and thus may explain the stability of microbial population size in mesocosm studies. While the current study did not show significant differences in prokaryotic cell numbers, the clone library data did suggest that the microbial diversity in filtered mesocosms was significantly reduced. This may be part of the ‘killing the winners’ phenomenon often observed in similar marine systems (Pernthaler, 2005). The percentage of dominant OHCB taxa, for example, A. borkumensis and related Oceanospirillales, were significantly reduced in the unfiltered mesocosms at all of the experimental sites, whereas the percentage of Alphaproteobacteria, for example, the members of Roseobacter spp., increased. Protozoan cell counts for day 25 were five to ten times higher in the unfiltered mesocosms from the Bangor and Helgoland sites. Observations made in prior studies indicate that ciliates are more likely to be abundant in experimental systems including microbial biomass aggregates such as emulsified oil (Gertler et al., 2010), which was present in mesocosms from all three locations. Because of their adaptations for grazing rather than interception feeding, ciliates predominantly feed on microorganisms in the biofilm, which may explain the reduced number of biofilm forming A. borkumensis-derived clones in the ‘grazing’-type mesocosms.

Implications for the development of bioremediation techniques

Although the current study was comprised of only three sampling sites located in European coastal waters, it nevertheless may provide vital insights into the development of novel oil spill mitigation techniques. Although our knowledge base on oil-degrading bacteria is vast, the variation in experimental design has complicated our ability to make unifying predictions about the significance of microbial community compositions in a range of geographical locations. This study has shown that a central and integral part of the oil-degrading consortia at all sites were organisms related with A. borkumensis. This is a common finding for oil degradation studies in Europe, and similar studies in East Asian waters have detected a number of closely related species exclusively from the Pacific Ocean, for example, Alcanivorax hongdengensis, Alcanivorax indicus and Alcanivorax pacificus. Comparable studies of Alcanivorax spp.-specific alkane hydroxylase genes isolated from the Atlantic Ocean and the Pacific Ocean have detected species such as A. venustensis, A. dieselolei and A. borkumensis (Wang et al., 2010ab). In contrast to the ubiquitous presence of primary oil-degrading microorganisms such as A. borkumensis, the abundance of other primary OHCB and secondary oil-degrading microorganisms such as C. spirillensus was highly variable. At the Bangor experimental site, two additional OHCB were detected in low numbers, indicating that these microorganisms could be location specific but possibly significant in the process of oil degradation. The detection of O. antarctica so far has only been reported at higher latitudes or in deep-sea water bodies at low temperatures; however, the relatively high incubation temperature of 16 °C as well as the initial temperature of sea water used for our experiments indicates that this organism is able to compete with other OHCB in northern temperate zones. This could imply that a temperature gradient may affect the composition of OHCB within oil-degrading microbial consortia.

This study has also detected two members of the Rhodobacteraceae that were found over the course of the experiment at all three sites. Although these microorganisms have been previously identified in oil-based communities, their importance in marine oil degradation is still unclear. We found that 58 of 64 OTUs detected in the clone libraries were site specific, which highlights the complexity of microbial ecology in marine oil degradation. Oil type and composition is considered to be an important factor in determining the composition of an oil-degrading consortia at a specific experimental site (McKew et al., 2007). Conversely, the experimental conditions and oil composition used in our study have demonstrated that while the indigenous microbial community at each site was highly variable, specific keystone species for the oil degradation process remained constant at all three experimental locations. This is supported by the results of both fingerprinting methods and clone library data and verified by multivariate statistics (PCO, permanova). Site-specific microbial consortia were detected at early and late stages of the experiment, yet fingerprint patterns of all three locations showed intense convergence during oil degradation. Although the primary colonization ofoil/water interfaces by OHCB such as A. borkumensis occurs naturally with a certain reliability, the complexity of consortia involved in secondary oil degradation following initial OHCB blooms and the logistics of sustaining them at the location of an oil spill make the use of artificially introduced bacterial cultures or engineered consortia for bioaugmentation purposes an entirely empirical enterprise. This study has shown location specificity of microbial communities dominated by OHCB at three sites to a certain degree, although we would hypothesize that site specificity also continues on a global scale. Thus, development of a universally applicable bioaugmentation system together with the corresponding production and storage of microbial consortia is highly impracticable and economically unviable. The alternative to such a system is the development of an inexpensive all-purpose universal biostimulation technique, which is designed to suit microbial consortia independently of both location and oil type. As key species for oil degradation are well known, such a technique could be designed to efficiently stimulate the growth of these bacteria without hindering the development of a location-specific consortium.

Conclusion

The current study investigated site specificity of oil-degrading microbial consortia through a standardized mesocosm experiment. Data suggest that while a few identical key species were detected in all sites and both microbial population size and dynamics showed high similarity, composition of the consortia was largely site specific and included variability in OHCB abundance. These results underline the complexity of marine oil-degrading microbial communities and cast further doubts on the necessity of bioaugmentation practices for successful bioremediation. The site specificity of microbial communities together with the universal abundance of primary oil-degrading microorganisms at all sampling sites suggests that biostimulation of these readily available, locally adapted OHCB would be more beneficial compared with less cost-efficient bioaugmentation approaches.

Acknowledgements

The authors would like to thank Ian Pritchard and Andy Beaumont of Bangor University's School of Ocean Sciences, Antje Wichels of Biologische Anstalt Helgoland, Angelika Arnscheidt of Helmholtz Center for Infection Research and the staff of both IAMC Messina and ECOSUD Italia for their assistance in this project. CG and PNG acknowledge the Centre for Integrated Research in the Rural Environment (CIRRE) and the EU FP7 Project ULIXES (266473). DJN was supported by the EU FP7 project BACSIN (211684).

Ancillary