Introduction

Sessile marine eukaryotes often harbour microbial biofilm communities on their surfaces. These microorganisms have primarily been of interest as potential sources of novel bioactive compounds (El-Gendy et al., 2008; Romanenko et al., 2008; Zhang et al., 2009). However, understanding the complex interactions of host–microbe associations has also become the focus of many microbial ecology studies. Such studies typically focus on the interactions of the microorganisms at one of three different levels, either: (i) within the microbial community; (ii) between the microbial community and the host; or (iii) between the microbial community and colonizing or grazing eukaryotes. There are several recent examples in the literature that address macro-algal bacterial interactions in this context. For example, microbial community interactions have been investigated using bacterial isolates from the green alga Ulva australis in experiments addressing competition in dual species biofilms and biofilm development (Mai-Prochnow et al., 2004; Rao et al., 2005). At the microbe–host level, it has been shown that for normal morphological growth of many green foliaceous macroalgae the presence of certain bacterial strains is essential (Nakanishi et al., 1999; Matsuo et al., 2005; Marshall et al., 2006). Finally, algal bacterial epiphytes have been shown to serve as settlement cues and deterrents for a range of eukaryotic larvae and spores (Dobretsov and Qian, 2002; Patel et al., 2003; Huggett et al., 2008). Cultured bacterial isolates were used to study these interactions.

In order to fully assess and explore such interactions, an improved understanding of the composition and dynamics of the surface community is required. Community fingerprinting methods, such as denaturing gradient gel electrophoresis (DGGE) and T-RFLP, allow for comparisons of many different replicates or samples to answer ecological questions concerning both the membership and stability of microbial communities. To understand the physical distribution and abundance of these microbial communities direct observations are required; fluorescence in situ hybridization (FISH) has been applied to study the bacterial communities associated with a number of marine hosts including sponges, corals and ascidians (Ainsworth et al., 2006; Hoffmann et al., 2006; Martinez-Garcia et al., 2007; Santiago-Vazquez et al., 2007; Neulinger et al., 2009), as well as to study the bacteria causing gall formation in the marine red algal genus Prionitis (Ashen and Goff, 2000). Wherever the application of FISH directly to algae is difficult because of intense background autofluorescence from algal cells, we previously successfully developed a catalyzed reporter deposition or catalyzed reporter deposition-fluorescence in situ hybridization (CARD-FISH) protocol to study the surface bacterial community of several marine macroalgae (Tujula et al., 2006).

Thus far, very limited data on the epiphytic communities on marine seaweeds is available. Here, we report on a comprehensive spatial and temporal analysis of the bacterial community associated with a green Ulvacean alga using a combination of DGGE and CARD-FISH. U. australis is found along the rocky shoreline of the Australian coast and belongs to the Ulvacean family of marine foliaceous macroalgae, which are distributed worldwide in intertidal and subtidal coastal environments (Womersley, 1996; Woolcott and King, 1999). We investigated the stability of the bacterial community by applying 16S rRNA gene DGGE bacterial community fingerprinting on multiple environmental samples of this alga, collected using a stratified design that allowed for spatial and temporal comparisons. This is the first study to employ such sampling strategy on a green alga and our results suggest that the epiphytic bacterial community is variable, yet contains members that are consistently associated with the host. We also identified the phylogenetic affiliation of members of the bacterial community and report on the presence of members of the Alphaproteobacteria, Gammaproteobacteria and the Bacteroidetes. Moreover, we determined the abundance of several bacterial groups on the algal surface, which revealed the dominance of Alphaproteobacteria, and in particular members of the Roseobacter clade, on the surface of U. australis.

Materials and methods

Sample collection

Ulva australis was harvested at low tide from the intertidal rock platform at Shark Point, Clovelly, NSW, Australia (33°91′S, 151°26′E). Sampling was conducted at 3 monthly intervals in 2004 to obtain representatives from different time points that corresponded to the four seasons. The water temperature for each sampling occasion was 23.3 °C in summer, 22.5 °C in autumn, 18.0 °C in winter and 19.0 °C in spring. On each occasion, four individual plants were collected from within each of three different tidal pools, located <20 m apart from one another (A–C) (n=12). Individual alga were collected into separate sterile plastic bags with approximately 500 ml of seawater from the site and placed on ice for transportation to the laboratory (approximately 0.5 h). Individuals were rinsed thrice with sterile seawater (filtered through a 0.22 μm filter and autoclaved) to remove debris and non-attached bacteria from the sample surface before further processing. Previous microscopic analysis has indicated that U. australis does not harbour endophytic bacteria (data not shown), and as such only surface associated bacteria should be analyzed here.

DNA extraction, 16S rRNA gene PCR, DGGE and sequencing

Nucleic acids were extracted from U. australis using the FastDNA spin kit for soil (QBiogene, Montreal, QC, Canada) using a standard protocol (Skovhus et al., 2004). For each algal sample, a sterile cork borer was used to cut five circular discs (1.5 cm in diameter, approximately 30% of the thallus length) from the fronds to ensure an equivalent amount (that is, surface area) was extracted. The remainder of the alga was fixed as described below for CARD-FISH analysis. On the first sampling occasion DNA extractions and subsequent PCR and DGGE analysis were carried out in triplicate to address variability on the level of the individual sample. A standard quantity of environmental DNA (approximately 20 ng) was used as template for PCR. PCR amplification and subsequent DGGE analysis of the bacterial 16S rRNA gene amplified with the primers 341F-GC and 907RC (Schafer et al., 2001), were carried out as described previously (Taylor et al., 2004), with two minor modifications. A denaturing gradient of 35–55% was used for DGGE. For data analysis the PRIMER program (v5.2.3) was used to generate Bray-Curtis similarity matrices from the presence–absence matrices of DGGE bands, and similarity dendrograms based on the group average. For the analysis of spatial variation between rock pools, all 12 individuals sampled from each time point were used. For the analysis of temporal variation over seasons, three individual samples were randomly selected from each time point. Selected DGGE bands were excised and sequenced using a standard protocol (Taylor et al., 2005). The TOPO TA cloning kit (Invitrogen, Carlsbad, CA, USA) was used for DGGE bands for which initial sequencing indicated the presence of multiple sequences, and up to 10 clones per band were sequenced. Wherever bands were detected across all samples, several bands from the corresponding vertical positions on the gel, but in different lanes, were also sequenced to confirm that they contained the same sequence (data not shown).

Nucleotide sequence accession numbers

The partial sequences of 16S rRNA genes obtained in this study are available from GenBank under accession numbers DQ229319–DQ229347 (see also Figure 3).

Phylogenetic analysis

The ARB software package (Ludwig et al., 2004) was used to carry out phylogenetic analysis. DGGE sequences and their nearest neighbours, determined by using the using seqmatch function with default parameters in the Ribosomal Database Project release 10 (http://rdp.cme.msu.edu/seqmatch/seqmatch_intro.jsp), were aligned using the SINA web aligner (http://www.arb-silva.de/aligner/) then imported into the ARB database. The phylogenetic trees were calculated using the maximum likelihood algorithm with near full-length 16S rRNA gene sequences of nearest neighbours (>1300 nt). To exclude the highly variable positions in the 16S rRNA gene a positional variability by parsimony mask was applied that resulted in the comparison of 1077 nucleotides in the tree calculations. Sequences of DGGE bands and other shorter sequences were added to the trees according to maximum parsimony criterion. An uncultured Crenarchaeote sp. (AJ535134) was used as an outgroup for the analysis. The robustness of the tree topology was tested by parsimony bootstrap analysis using 1000 resamplings.

CARD-FISH

All algal samples were fixed in paraformaldehyde (2%) and stored in 1 × PBS:ethanol (4:6) at −20 °C using a standard protocol (Amann et al., 2001). The three samples from each season assessed in the above seasonal DGGE comparison were analyzed using an adapted CARD-FISH protocol optimized for algae (Tujula et al., 2006). Probes were chosen to target bacterial groups on U. australis based on the DGGE sequences obtained, which although not an exhaustive exploration of the diversity present, should represent the most abundant groups within the community. After matching against the SILVA database in ARB (Pruesse et al., 2007) to assess the likelihood of a match to sequences with high similarity to DGGE sequences obtained in this study, we applied the following probes labeled with horseradish peroxidase (Thermo Electron Corporation, Bremen, Germany): EUBmix (EUB338-i, ii and iii) (Daims et al., 1999) to target the bacterial kingdom, ALF968 (Glockner et al., 1999) which targets the Alphaproteobacteria, GAM42a (Manz et al., 1992) for the Gammaproteobacteria with unlabeled competitor probes GAM42a_C1033 (Yeates et al., 2003) and BET42a (Manz et al., 1992), CF319a (Manz et al., 1996) for the Bacteroidetes, ROS537 (Eilers et al., 2001) to target the Roseobacter clade and PSU730 (Huggett et al., 2008) for the Pseudoalteromonas genus. The three summer samples had all six probes applied in order to determine the relative abundance of each group. The probes EUBmix, ALF968 and ROS537 were additionally applied to three autumn, winter and spring representatives to assess the variability of these groups between seasons. Hybridization and wash buffers were prepared according to Pernthaler and Pernthaler (2007). Tyramide-labeled Alexa546 (Invitrogen) was used as the reporter molecule for signal amplification. Samples were counterstained with SYBR Green II (2 × concentration, Invitrogen). For each sample probed, 20 random images were collected by confocal laser scanning microscopy and quantified using the Image J software program (http://rsb.info.nih.gov/ij) as outlined previously (Tujula et al., 2006).

Results

Spatial and seasonal variability in 16S rRNA gene DGGE bacterial community fingerprints

A total of 48 environmental samples of U. australis collected as 12 individuals at 3 monthly intervals were processed. Cluster analysis of DGGE profiles showed that the variability between the communities of individual hosts was generally high (Figure 1). The range of variability between individuals was lowest for the autumn time point (22%) and was highest in summer (40%), followed by winter (35%) and spring (33%). In terms of spatial variation, cluster analysis indicated that generally the differences in banding patterns between individuals collected from different tidal pools were no greater than the differences between individuals from within a tidal pool (see Figure 1). Thus, we randomly selected three individuals from each seasonal sampling point to assess temporal variation. With the exception of one individual from winter, the community banding profiles of samples from a given time point were more similar to each other, than to samples from another season (Figure 2b). This cluster analysis also revealed that the bacterial communities of samples collected during different seasons still share approximately 60% of bands.

Figure 1
figure 1

16S rRNA gene-based similarity dendrogram showing the relationships among the microbial communities of different individuals of Ulva australis collected from three different tidal pools (A, B, C) over four seasonal time points. This dendrogram was generated using the PRIMER program.

Full size image
Figure 2
figure 2

Seasonal comparison of microbial communities of Ulva australis. (a) 16S rRNA gene DGGE gel representing individuals (1–3) from summer (S), autumn (A), winter (W) and spring (R). M=marker. Numbers indicate which bands were sequenced for phylogenetic analysis, and correspond to band numbers in Figure 3. (b) 16S rRNA gene-based cluster diagram representing the data from (a). This dendrogram was generated using the PRIMER program.

Full size image

Phylogenetic analysis of excised 16S rRNA gene DGGE bands

Selected bands from the DGGE gels were excised and sequenced to identify the phylogenetic affiliations of representative members of the surface community. A total of 34 sequences were obtained, from which 28 distinct sequence types were subjected to phylogenetic analysis (Figure 3). The majority of sequences (15) obtained were affiliated with the Alphaproteobacteria and belonged to the families Rhodobacteraceae, Phyllobacteriaceae and Sphingomondaceae. It is to be noted that a number (7) of related sequences (89–97%) obtained clustered within the Roseobacter clade. The second most represented phylogenetic group of sequences belonged to the Bacteroidetes phylum (7). Of these, the majority belonged to the Sphingobacteriales (6), however, one sequence fell within the Flavobacteriales. We also obtained a sequence belonging to the Gammaproteobacteria (1), an unclassified Protoebacteria sequence (1) and several unaffiliated sequences (4) that were related (87–93% identity) to clone sequences from a carbonate chimney in the Lost City hydrothermal field and clones within the candidate bacterial phylum OP11.

Figure 3
figure 3

Maximum likelihood tree showing the phylogenetic position of Ulva australis surface bacterial 16S rRNA gene bands excised from DGGE gels (shown in bold). The tree was generated in ARB by comparing 1077 nucleotides of the 16S rRNA gene, with shorter DGGE sequences added afterwards using the Parsimony insertion tool. Closed circles at nodes indicate >90% bootstrap support by parsimony analysis, using 1000 resamplings. An uncultured Crenarchaeote sp. (AJ535134) was used as an outgroup for the analysis but is not shown in the tree. The scale bar indicates 0.1 nucleotide changes (10%) per nucleotide position. * denotes a sequence from a DGGE band that was detected in all samples; letters after a number indicate where multiple clones were sequenced from one band.

Full size image

It is interesting to note that sequences from five DGGE bands were detected in all samples analyzed, although one represented three different sequences (band 19). The sequences from three bands were affiliated with the Alphaproteobacteria and one within the Bacteroidetes. Specifically, DGGE bands 25 and 4 were closely related (>99%) to Alphaproteobacteria clone sequences earlier obtained from two seaweed surfaces, as well as uncultured clone sequences (>97%) from healthy coral tissue. Hellea balneolensis was the nearest cultured relative of DGGE band 18, which was also related to an Alphaproteobacteria clone (>94%) derived from the microbial community of U. australis (Longford et al., 2007). The DGGE band within the Bacteroidetes (DGGE band 1) was related (97%) to clones derived from U. australis (Longford et al., 2007) and was almost identical (99.6%) to the ocean sediment isolate Krokinobacter diaphorus (AB198089) (Figure 3).

In situ abundance of epiphytic bacterial groups on the surface of U. australis

In order to gain an understanding of the physical representation and abundance of bacterial groups identified by DGGE on the host, CARD-FISH was applied. A hierarchical approach was used when selecting and applying probes to these samples. Initially, the domain Bacteria was targeted and found to account for an average of 90% (±1.6) of the cells in the microbial community on the seaweed surface (Figure 4a), visually comprising a diverse range of cell morphologies (see Figure 5). Many bacteria were observed in the form of microcolonies (such as in Figure 5f) often along the algal intercellular cell wall depressions, and were comprised of a variety of cell morphologies (data not shown).

Figure 4
figure 4

Percentage of specific bacterial groups enumerated by CARD-FISH of the total number of SYBR Green II-stained cells on the surface of Ulva australis. Error bars indicate standard deviation. (a) Samples collected in summer 2004 (n=3) were hybridized with the six probes listed in the key above, targeting Bacteria (EUBi-iii), Alphaproteobacteria (ALF968), Roseobacter (ROS537), Bacteroidetes (CF319a), Gammaproteobacteria (GAM42a) and Pseudoalteromonas (PSU730). (b) Samples collected in autumn (n=3), winter (n=3) and spring (n=3) 2004 were further hybridized with probes targeting Bacteria, Alphaproteobacteria and Roseobacter.

Full size image
Figure 5
figure 5

Confocal laser scanning micrographs of bacterial cells on U. australis visualized by CARD-FISH using horseradish peroxidase probes targeting the Alphaproteobacteria (a–d), Bacteroidetes (e) and Roseobacter (f). Alexa546 was used as the reporter signal and SYBR Green II as a counter stain, the probe-conferred signal is presented in red and SYBR Green II counterstaining is presented in green. The scale bar on all panels is 10 μm.

Full size image

Quantification of the Alphaproteobacteria, Gammaproteobacteria and Bacteroidetes showed that the Alphaproteobacteria comprised on average 70% (±1.0) of the cells in the microbial community (Figure 4a). Four distinct alphaproteobacterial cell morphotypes were noted (Figures 5b–e): (i) thin rods (0.5 μm in diameter and 3 μm in length); (ii) filamentous cell chains (1.5 μm diameter, 10–15 μm in length); (iii) diploid large rods (1.5 μm diameter, 3 μm long, combined length 6 μm); and (iv) coccibacilli (approximately 1.5 μm in diameter).

The average proportion of the Bacteroidetes cells was 13% (±7.4) (Figure 4a). The majority of cells were thin and rod shaped in appearance, and can be seen in association with microcolonies of other bacteria (Figure 5f). Very few Gammaproteobacteria cells (<1%) were detected in this study (Figure 4a).

At the final level in our probe hierarchical analysis, a group from each of the Alphaproteobacteria and Gammaproteobacteria was quantified. The marine-alphaproteobacterial lineage, which includes Roseobacter, was selected as it was represented by many of the DGGE band sequences. The Roseobacter population was found to have a relative abundance of 12% (±7.3) (Figure 4a) and a sole coccobacilli morphotype was observed with cells often noted together in aggregates or as microcolonies (Figure 5f). The gammaproteobacterial genus Pseudoalteromonas was chosen, even though no such DGGE sequences were obtained, because several studies have reported a high proportion of culturable isolates of this genus from the surface of Ulvacean algae (Egan et al., 2000, 2001; Dobretsov and Qian, 2002; Patel et al., 2003). However, Pseudoalteromonas cells were not readily detected.

Following the observed seasonal variation in the bacterial community composition (see Figure 2), the three individuals from each of the seasonal sampling points were further investigated to assess the variability of the Alphaproteobacteria and the Roseobacter clade. Across all four seasons, the abundance of Bacteria within the epiphytic community was 91.9% (±1.9). Alphaproteobacteria and Roseobacter accounted for 71.0% (±4.9) and 19.1% (±9.4), respectively, although the abundance of Roseobacter exhibited highest variability and was most abundant in the winter samples (Figure 4b).

Discussion

This is the first study to address the spatial and temporal variability in the bacterial community associated with a cosmopolitan Ulvacean algae, employing both 16S rRNA gene DGGE and CARD-FISH analysis. DGGE data available on other marine macroalgae derived from rpoB based DGGE show differences in banding patterns among individuals of up to 16% for Amphiroa anceps, 30% for Corallina officinalis and 30% for Delisea pulchra (Huggett et al., 2006). Studies of sponge-associated bacteria based on DGGE analysis of the 16S rRNA gene reported differences in DGGE banding patterns of <30%, even for samples of the same species collected over a 500 km range (Taylor et al., 2005). Here, we report differences of up to 40% when comparing individuals of U. australis from within the same, and between three different, tidal pools separated by <20 m, which indicates that the phylogenetic composition of this alga is perhaps more variable than other host-associated communities from similar habitats.

In our temporal comparison, community variability correlated with the season of sample collection. Seasonal differences have been observed in non-DGGE based investigations of epiphytic microbial communities of algae and have been attributed to the growth state of the plant (Laycock, 1974; Sieburth and Tootle, 1981). Leaf age is considered to be a significant inherent source of variation in spatial and temporal assessments (Baker, 1998), and this has been demonstrated recently in a study of the brown alga Laminaria saccharina, wherein DGGE profiles of the young meristem and cauloid sections from different plants were more similar to each other than the ageing phyloid section of the same plant (Staufenberger et al., 2008). It is possible that leaf age also contributes to spatial and temporal variability observed in this study, as attempts to collect algal samples of similar size, a crude estimator of age, proved difficult in the field. However, the data reported here indicates that despite levels of variability, the epiphytic community includes a sub-population of bacteria that are consistently associated with U. australis, as reflected by the DGGE bands present in all samples. The consistent detection of these sequences indicates that this sub-population may have an important role in the function of the bacterial community on U. australis. Similar observations of specificity have recently been reported for bacterial assemblages associated with marine diatoms (Grossart et al., 2005), sponges (Taylor et al., 2005; Lee et al., 2009) and corals (Webster and Bourne, 2007), suggesting that studies of such microbial communities will increasingly add to our understanding of marine bacterial sessile host interactions.

The majority of DGGE sequences obtained in this investigation were affiliated with the Alphaproteobacteria and the Bacteroidetes, including bands detected in all samples, indicating that representatives of these broad phylogenetic groups constitute the key members of the epiphytic community. A prevalence of sequences from these two bacterial groups has been reported in a DGGE based study of phytoplankton (Jasti et al., 2005), and on brackish submerged macrophytes (Hempel et al., 2008). In addition to these two groups, Gammaproteobacteria and unaffiliated sequences similar to the candidate bacterial phylum OP11, were identified in this study. A 16S rRNA gene clone library of the epiphytic bacterial community associated with U. australis and the red alga Deslisea pulchra (Longford et al., 2007) also contained sequences belonging to these bacterial groups and several of our DGGE bands were expectedly related to these clones. These broad phylogenetic groups, with the exception of the OP11-like sequences, have also been detected in molecular-based community analyses of bacterial assemblages associated with the tropical green macro-alga Caulerpa taxifolia (Meusnier et al., 2001), filamentous green algae of sphagnum bogs (Fisher et al., 1998), marine diatoms (Grossart et al., 2005) and phytoplankton (Jasti et al., 2005), and hence appear to emerge as important associates of algae in general.

The confocal laser scanning micrographs of the bacterial community of U. australis (Figure 5) complement the early scanning electron micrographs of U. lactuca thalli (Sieburth, 1975) that have aptly been described as luxuriant microbial gardens. By using CARD-FISH analysis we found that bacteria accounted for more than 90% of the cells in the microbial community over all four seasons. This is relatively high in comparison to the average of 56% determined in a review of FISH studies in aquatic environments (Bouvier and del Giorgio, 2003) and could be a reflection of the importance of bacteria for the normal morphological growth of Ulva (Provasoli and Pintner, 1980; Nakanishi et al., 1996; Matsuo et al., 2003). The larger non-hybridized cells observed frequently on the surface of U. australis are possibly diatoms and microalgae, which is in agreement with scanning electron micrographs-based observations of microfouling of small diatoms and blue-green algae on seaweed surfaces (Sieburth, 1975; Provasoli and Pintner, 1980; Dobretsov and Qian, 2002).

Many bacteria were observed in microcolonies (as can be seen in Figure 5f) in association with the algal intercellular cell-wall junctions (data not shown). This observation has also been made for kelp species of the genera Laminaria and Ecklonia (Linley et al., 1981; Koop et al., 1982; Corre and Prieur, 1990) and is thought to be related to the accumulation of nutrients in the depressions between cells on the algal surface. The intercellular spaces of an unclassified Ulva sp. have been found to contain packed layers of polysaccharides containing xylose and glucose (Bobin-Dubigeon et al., 1997), which most likely act as carbon source for bacterial microcolonies observed in these areas. The formation of microcolonies may also protect bacteria from potential stresses within shallow tidal pools such as desiccation, osmotic stress and UV radiation, as has been shown for bacteria on terrestrial leaf surfaces (Monier and Lindow, 2003).

The selected probes used for CARD-FISH analysis were based on the DGGE sequence information obtained in this study. Both methods correlated in identifying Alphaproteobacteria as the most abundant group on the surface of U. australis. It is difficult to relate the dominance observed in this study to particular functional phenotypes, because the Alphaproteobacteria are morphologically and metabolically extremely diverse. However, recent evidence indicates that the Alphaproteobacteria contribute significantly to dimethylsulfoniopropionate (DMSP) assimilation in oceans and have an important role in global sulphur cycling (Malmstrom et al., 2004). Furthermore, DMSP is an osmolyte produced by phytoplankton and algae, including Ulva sp., which has been demonstrated to provide protection to bacteria during salt stress (Pichereau et al., 1998; Cosquer et al., 1999). Thus, we speculate that the high proportion of Alphaproteobacteria may be linked to DMSP utilization.

The Roseobacter clade is an important marine alphaproteobacterial lineage (for review see Brinkhoff et al. (2008)) that includes many species highly efficient in DMSP degradation and assimilation (Moran et al., 2003; Malmstrom et al., 2004). Bacteria within this clade are known to be ubiquitous and rapid colonizers of surfaces in coastal environments (Dang and Lovell, 2000) and have been enumerated as the most abundant group within the bacterial assemblages associated with some marine algal cultures and phytoplankton blooms in nature (Gonzalez et al., 2000; Riemann et al., 2000; Alavi et al., 2001). Given that many of the DGGE sequences fell within the Roseobacter clade, this group was selected for a detailed quantification on the surface of U. australis. The seasonal CARD-FISH data support the idea that Roseobacter species are an integral part of the epiphytic community of U. australis, in agreement with the finding that an epiphytic Roseobacter isolate was able to out-compete all other strains trialed in dual species biofilm competition experiments with bacterial isolates from U. australis (Rao et al., 2005).

Members of the Bacteroidetes were the second most abundant bacteria on the algal surface. The abundance of Bacteroidetes on the surface of environmental samples of U. australis in the present study is approximately 13% of surface associated bacterial cells. As Bacteroidetes bacteria have been found to contribute 15–30% of the DMSP assimilation in the North Atlantic Ocean and the Gulf of Mexcio (Malmstrom et al., 2004), it is possible that the relatively high abundance of this group may also be linked to DMSP production by the alga. Bacteroidetes bacteria have also been linked with the maintenance of normal thallus structure in the green alga Monostroma oxyspermum (Matsuo et al., 2003), although bacteria from a range of phyla including the Proteobacteria have also been shown to induce normal morphology in the green alga Ulva linza (Marshall et al., 2006). Six of the seven Bacteroidetes sequences are closely related to the Saprospiraceae family, strains of which have been isolated from seawater (Lee, 2007; Oh et al., 2009) and algae (Hosoya et al., 2006). Saprospiraceae sequences have also been detected in a previous study of U. australis and the red alga Delisea pulchra (Longford et al., 2007).

In this investigation we obtained one gammaproteobacterial DGGE sequence and using CARD-FISH quantified the abundance of this group to be <1%. This low in situ abundance appeared to be in contrast with numerous reports of a high proportion of gammaproteobacterial species isolated from the surface of Ulvacean algae (Egan et al., 2000, 2001; Vairappan and Suzuki, 2000; Dobretsov and Qian, 2002; Patel et al., 2003) and highlights the culturing bias frequently observed in microbial ecological studies (Hugenholtz, 2002). However, recent investigations suggest that organisms having a low in situ abundance should not be disregarded in terms of their function within the community. For example, members of the genera Pseudoalteromonas, Vibrio and Shewanella present at an abundance of <1% of the epiphytic community of the algae C. offencialis and A. anceps significantly contribute to induce the settlement of sea urchin larvae (Huggett et al., 2008).

Concluding remarks

The two community profiling techniques DGGE and CARD-FISH analysis employed in this study identified Alphaproteobacteria as the dominant bacterial group, which together with the Bacteroidetes constitute the majority of bacterial cells in the epiphytic community on the surface of U. australis. The temporal and spatial comparisons carried out have revealed that the bacterial community associated with U. australis is variable both between individual plants and across different seasons. However, there remains a sub-population of bacteria that are found consistently across space and time, again affiliated with the Alphaproteobacteria and the Bacteroidetes suggesting that they are likely to have an important role in the function of this bacterial community. The presence of stable sub-populations within host associated bacterial communities is a feature that is likely to be germane to the ecology of host-associated microbial communities.