|Home | About | Journals | Submit | Contact Us | Français|
Bacteria that produce inhibitory compounds on the surface of marine algae are thought to contribute to the defense of the host plant against colonization of fouling organisms. However, the number of bacterial cells necessary to defend against fouling on the plant surface is not known. Pseudoalteromonas tunicata and Phaeobacter sp. strain 2.10 (formerly Roseobacter gallaeciensis) are marine bacteria often found in association with the alga Ulva australis and produce a range of extracellular inhibitory compounds against common fouling organisms. P. tunicata and Phaeobacter sp. strain 2.10 biofilms with cell densities ranging from 102 to 108 cells cm−2 were established on polystyrene petri dishes. Attachment and settlement assays were performed with marine fungi (uncharacterized isolates from U. australis), marine bacteria (Pseudoalteromonas gracilis, Alteromonas sp., and Cellulophaga fucicola), invertebrate larvae (Bugula neritina), and algal spores (Polysiphonia sp.) and gametes (U. australis). Remarkably low cell densities (102 to 103 cells cm−2) of P. tunicata were effective in preventing settlement of algal spores and marine fungi in petri dishes. P. tunicata also prevented settlement of invertebrate larvae at densities of 104 to 105 cells cm−2. Similarly, low cell densities (103 to 104cells cm−2) of Phaeobacter sp. strain 2.10 had antilarval and antibacterial activity. Previously, it has been shown that abundance of P. tunicata on marine eukaryotic hosts is low (<1 × 103 cells cm−2) (T. L. Skovhus et al., Appl. Environ. Microbiol. 70:2373-2382, 2004). Despite such low numbers of P. tunicata on U. australis in situ, our data suggest that P. tunicata and Phaeobacter sp. strain 2.10 are present in sufficient quantities on the plant to inhibit fouling organisms. This strongly supports the hypothesis that P. tunicata and Phaeobacter sp. strain 2.10 can play a role in defense against fouling on U. australis at cell densities that commonly occur in situ.
Surfaces never remain pristine in the marine environment but become quickly colonized by a film of marine bacteria. Such biofilms serve as cues that modify the behavior of settling invertebrate larvae and algal cells (12, 73) and significantly influence the final composition of the biofouling community. These cues can be either positive (11, 42, 43, 49, 57, 69) or negative (19, 33, 50), depending on the species of fouling organism and bacteria concerned. Thus, biofilms play an important role in the development of fouling communities on marine surfaces.
As well as being a common fouling organism, the intertidal green alga Ulva australis is itself susceptible to fouling because it is sessile and restricted to the photic zone, where conditions for fouling organisms are optimal (13). Many seaweeds have evolved efficient strategies to combat epibiosis (14, 17, 41), but these antifouling defenses can be energetically costly (72), and it has been suggested that epibiotic bacteria living on the surface of the alga may provide microbial defense (3, 18, 19, 33). Such interactions are not uncommon in the marine environment (16, 24, 25, 28). A number of inhibitory bacteria have been isolated from U. australis, and one of the best characterized of these is Pseudoalteromonas tunicata (18, 20, 32). This bacterium produces a diverse range of biologically active compounds that specifically target marine fouling organisms (19, 23, 30, 32, 33, 38) and gram-negative and gram-positive bacteria from a range of environments (48). Phaeobacter sp. strain 2.10 (formerly Roseobacter gallaeciensis) is also frequently isolated from the surface of U. australis and has known antibacterial activity (4, 58, 61). Members of the Roseobacter clade are frequently associated with algae (8) and can comprise up to 25% of microbial communities in coastal environments (71).
Molecular investigations based on real-time quantitative PCR have shown that while the genus Pseudoalteromonas is common throughout the marine environment, P. tunicata mostly inhabits living surfaces that are relatively free from fouling such as green algae (Ulva lactuca and Ulvaria fusca) and tunicates (Ciona intestinalis) (65) and has low in situ density (<1 × 103 cells cm−2) (66). Studies based on a method combining catalyzed reporter deposition with fluorescence in situ hybridization suggested that Phaeobacter sp. strain 2.10 may be present in higher numbers, as the genus Roseobacter comprised 12% of the epiphytic bacterial community on U. australis (unpublished data).
Although the inhibitory effects of P. tunicata against a range of fouling organisms in the laboratory are well established (18, 31), the observation that P. tunicata is present at such low densities raises the question as to whether it could in fact have antifouling activity at these densities in the marine environment. In this study, we tested the density dependence of antifouling activity of P. tunicata and Phaeobacter sp. strain 2.10 biofilms and show that they have inhibitory effects at ecologically relevant densities. Furthermore, the inhibition of algal spores and larval settlement by P. tunicata biofilms is shown to be due to the production of antifouling compounds, as mutants defective in the production of the extracellular inhibitors do not display inhibitory activity. Inhibition of settlement and attachment by Phaeobacter sp. strain 2.10 biofilms is through an unknown mechanism. However, Phaeobacter sp. strain 2.10 produces the density-dependent signal molecules acyl homoserine lactones (AHLs), which are also settlement cues for Ulva sp. spores (40). This suggests that AHLs may play a role in mediating the relationship between U. australis and quorum-sensing bacterial epiphytes capable of defending the alga. Our data support the hypothesis that P. tunicata and Phaeobacter sp. strain 2.10 play a role in defending U. australis against fouling in situ.
Bacterial strains were isolated from the surface of U. australis as described previously by Rao et al. (58). The organisms selected for this study were P. tunicata, Phaeobacter sp. strain 2.10, P. gracilis, Alteromonas sp., Cellulophaga fucicola; as controls, P. tunicata mutants defective in the production of the antibacterial protein (AlpP) (47), all inhibitory compounds (WmpR) (17), and the antifungal compound (FM3) (20) were used. Cultures were stored at −80°C in 50% (vol/vol) glycerol in VNSS medium (52) and maintained on VNSS agar plates. For visualization of bacterial cells using epifluorescence microscopy, P. tunicata, P. gracilis, Alteromonas sp., C. fucicola, and Phaeobacter sp. strain 2.10 were labeled with either DsRed or green fluorescent protein (GFP) color tag as described previously by Rao et al. (58).
Bacteria were cultured for 24 h at 25°C in VNSS broth for preparation of inocula. Cells were harvested by spinning down the culture and resuspending the pellet in seawater. The cell concentration was estimated by counting using epifluorescence microscopy and a hemocytometer and adjusted by dilution with seawater to the desired end concentration. Biofilms were established by inoculation from overnight precultures into 3.6-cm petri dishes containing 3 ml of 10% VNSS medium (diluted in seawater) and incubated at 23°C for 24 h. Different densities of cells were inoculated so that the final density of attached cells on the surface of the petri dishes ranged from 102 to 108cells cm−2. After 24 h of incubation, growth medium was discarded and established biofilms were rinsed three times with sterile filtered seawater and incubated in fresh sterile seawater before the bioassays were conducted. Numbers of attached cells were determined by counting the number of DsRed- or GFP-labeled cells with epifluorescence microscopy and an eyepiece grid. Twenty-four-hour-old biofilms were used for antifouling assays as older biofilms undergo a certain amount of detachment and sloughing, which affected cell densities. The biofilm densities were measured after the assays to establish whether the density had changed. Preliminary experiments indicated that cell densities in biofilms did not increase significantly over the course of the experiment, particularly for the lower densities. For all assays, three experiments were carried out with at least four replicates in each treatment.
The effects of different cell densities of bacteria on attachment and survival of algal propagules were assessed by exposing gametes or spores directly to monoculture biofilms of P. tunicata and Phaeobacter sp. strain 2.10 at densities ranging from 102 to 108 cells cm−2. Sterile seawater and P. tunicata WmpR mutant biofilms established at 106cells cm−2 served as controls for P. tunicata biofilms. Sterile seawater was the control for Phaeobacter sp. strain 2.10.
U. australis gamete and Polysiphonia sp. spore bioassays were set up as described by Egan et al. (19). U. australis gamete settlement was assessed after 5 days using an inverted light microscope (Zeiss). Counts were conducted in 10 fields of view using a ×40 magnification, and settlement was compared to that in controls. Polysiphonia sp. spore settlement and subsequent development were assessed after 24 h, the numbers of settled (i.e., attached) and unsettled spores were counted using a dissecting microscope, and the percentage of settlement was determined.
P. tunicata and Phaeobacter sp. strain 2.10 were screened for AHL production using a streak assay and the AHL reporter strain, Agrobacterium tumefaciens A136. Briefly, plates were poured with one medium and set, and half of the medium was aseptically removed and replaced with a second medium. This was done because P. tunicata, Phaeobacter sp. strain 2.10, and A. tumefaciens A136 require different growth media. A. tumefaciens A136 was grown on LB5 plates (2) supplemented with tetracycline (4.5 μg/ml), spectinomycin (50 μg/ml), and X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside [50 μg/ml]). Marine strains grew on marine broth 2216 supplemented with 1.5% agar.
Following a positive result in the AHL streak assay, AHLs were further characterized with thin-layer chromatography (TLC) plates overlaid with A. tumefaciens A136. To extract AHLs from P. tunicata and Phaeobacter sp. 2.10, the strains were grown to stationary phase in 10 ml of marine broth 2216. These 10-ml cultures were sterile filtrated (0.22 μm), and the supernatants were retained. AHLs were extracted from the supernatants as described by Ravn et al. (60), the AHL extracts were resuspended in 100 μl of acidified ethyl acetate, and 40 to 80 μl of each sample was applied to C18 TLC plates (TLC aluminum sheets, 10 by 10 cm2, Rp-18 F254 s,1.05559 [Merck 64271]) and developed in 60:40 methanol-Milli-Q water as described by Shaw et al. (63). A lawn of A. tumefaciens A136 in AB medium was overlaid on the TLC plate as described by Ravn et al. (60). AHLs were compared against commercial standards 3-oxo-hexanoyl homoserine lactone (OHHL), N-oxo-octanoyl-l-homoserine lactone (OOHL), and N-octanoyl-homoserine lactone (OHL), and Rf values were calculated (56). AHLs were identified based on Rf values and the shape of spots.
To determine if AHLs were involved in enhancing settlement of U. australis gametes, an assay was conducted with different concentrations of the C8-AHL, OOHL. OOHL was used because unsubstituted C7-AHL is not commercially available and AHLs with a carbon chain length of <6 are freely diffusible in water and those with a length of >10 have markedly lower solubility (55). OOHL was suspended in a 1% agarose-distilled water support matrix as described in reference 70 at concentrations of 5, 10, 20, and 50 μM, respectively. A control of 50 μM methanol in 1% agarose was used. Gamete attachment was assessed after 1 h, using an inverted light microscope (Zeiss). Gametes were stained with crystal violet, counts were conducted in 10 fields of view using a ×40 magnification, and settlement was compared to that in controls.
P. tunicata and Phaeobacter sp. strain 2.10 at a range of densities were tested using standard settlement assays of larvae of the bryozoan Bugula neritina (7). Adult broodstocks of B. neretina were collected from pilings at Rose Bay, Sydney, Australia (33°87′52”S, 151°25′56”E) and larvae were obtained as described by de Nys et al. (13) and Bryan et al. (7). Only newly released larvae were included in the bioassay (i.e., within 15 min of release). Biofilms were established as described above, petri dishes were rinsed three times with 2 ml of sterile seawater, and then 3 ml of filtered seawater containing approximately 20 larvae was added to each petri dish and incubated at 25°C for 2 days. Larvae were counted under a dissecting microscope, and the percentage of settlement was determined. Control petri dishes contained either sterile seawater alone (for both P. tunicata and Phaeobacter sp. strain 2.10) or the P. tunicata WmpR mutant's biofilm with cell densities of 106 cells ml−1 (control for P. tunicata).
Both yeast and filamentous fungi were used to assess the antifungal activity of P. tunicata and Phaeobacter sp. strain 2.10. Unidentified marine yeast (Y1, Y2, and Y3) and a filamentous fungal (Y4) strain previously isolated from U. australis (21) were used as target strains. Fungal strains were maintained on VNSS plates and inoculated into VNSS broth. Bacterial biofilms were established as described earlier, petri dishes were rinsed three times with 2 ml of sterile seawater, and 2 ml of medium (10% VNSS-90% seawater) containing 105cells ml−1 of fungi was added and incubated for 48 h. The fungi were stained with Syto 59, and the percentage of fungal surface cover was compared to that of the original inoculum by epifluorescence microscopy. Biofilms of P. tunicata FM3 and WmpR mutant cells established at 106 cells cm−2 served as controls. Counts were done on 10 fields of view using a ×40 magnification.
Marine strains isolated from U. australis were used as target strains to test for antibacterial activity of P. tunicata and Phaeobacter sp. strain 2.10 biofilms established on plastic surfaces. The bacterial challenge strains consisted of C. fucicola, Alteromonas sp., Phaeobacter sp. strain 2.10, P. tunicata, and P. gracilis. Biofilms of P. tunicata, and Phaeobacter sp. strain 2.10 were established and challenged with test bacteria added to the biofilms at 106 cells ml−1 and incubated for 48 h. P. tunicata AlpP and WmpR mutant biofilms and culture media were used as controls for P. tunicata, while culture medium was the control for Phaeobacter sp. strain 2.10. DsRed-labeled challenge strains were readily distinguished from GFP-labeled biofilm bacteria by epifluorescence microscopy. Counts were done on 10 fields of view using a ×10 magnification.
One-way analysis of variance (ANOVA) followed by Tukey's pairwise comparisons were used to compare the settlement of spores or larvae and attachment of bacteria or fungi in response to various densities of P. tunicata and R. gallaeciensis. ANOVA assumptions of normality and heterogeneity of variance of the data were checked. Tests were conducted with Systat 10 (SPSS).
P. tunicata grown as biofilms on plastic surfaces was inhibitory to Polysiphonia sp. spore settlement at cell numbers as low as 102 cells cm−2. At cell densities of 103 cells cm−2, inhibition was more than 90% (Fig. (Fig.1;1; ANOVA, F8, 36 = 862.70, P < 0.001). At higher cell densities (106 to 108cells cm−2), P. tunicata biofilms lysed Polysiphonia sp. spores.
U. australis gametes were less sensitive to inhibition by P. tunicata, and relatively higher cell densities were required to inhibit its settlement. Cell densities of 106 cells cm−2 were required for inhibition of U. australis gamete settlement to more than 90%; however, cell densities as low as 104 cells cm−2 showed a significant inhibitory effect (Fig. (Fig.2;2; ANOVA, F8, 36 = 49.27, P < 0.001).
In contrast, Phaeobacter sp. strain 2.10 biofilms grown on plastic plates had no effect on Polysiphonia sp. spore settlement at any of the densities tested (data not shown). This is in contrast to another Phaeobacter isolate, which produces the inhibitory compound tropodithietic acid that is toxic toward microalgae (4).
Phaeobacter sp. strain 2.10 did not inhibit U. australis gamete settlement: in fact, it mildly stimulated spore settlement at high cell densities (Fig. (Fig.3;3; ANOVA, F7, 32 = 4.00, P < 0.001). Joint et al. (39, 40) have shown that the bacterial quorum-sensing molecules, AHLs, enhance Ulva sp. zoospore settlement. Therefore, we tested Phaeobacter sp. strain 2.10 for production of AHLs and detected one AHL by using TLC plates overlaid with the AHL reporter strain A. tumefaciens A136. The AHL produced by Phaeobacter sp. strain 2.10 is tentatively identified as an unsubstituted C7-AHL with an Rf value of 0.38. Many species belonging to the Roseobacter clade have recently been identified as producing AHLs (5, 6); however, the role of quorum sensing in any of these species is yet to be elucidated. Because Ulva sp. spores are chemotactic to AHLs, this suggests that quorum-sensing species have a beneficial relationship with Ulva spp. While many quorum-sensing species are pathogens, some quorum-sensing bacteria act as biocontrol strains for terrestrial plants (9, 45).
In order to examine in more detail whether AHLs play a role in U. australis gamete settlement, a settlement assay was conducted with different concentrations of the AHL because OOHL as unsubstituted C7-AHL is not commercially available. The settlement of U. australis gametes generally increased with increasing concentrations of AHLs (data not shown). However, concentrations of AHLs produced by Phaeobacter sp. strain 2.10 in biofilms remain to be determined. The reasons for enhanced gamete settlement in the presence of AHLs are unclear, but it has been suggested that such responses may facilitate a close relationship between germinating spores and beneficial bacteria (40). The production of bioactives against other fouling organisms by Phaeobacter sp. strain 2.10 may be characteristic of such beneficial bacteria.
Biofilms of P. tunicata on plastic surfaces also inhibited settlement of B. neretina larvae at relatively high densities (105 to 106 cells per cm−2) (Fig. (Fig.4;4; ANOVA, F8, 36 = 166.09, P < 0.001). P. tunicata has previously been shown to exhibit antilarval effects at high densities on polystyrene plates (33). Previous studies have demonstrated that members of the genus Pseudoalteromonas can both inhibit (16, 31, 46) and induce (47, 53) larval settlement.
Phaeobacter sp. strain 2.10 biofilms inhibited bryozoan larval settlement at densities as low as 104 cells cm−2 (Fig. (Fig.5;5; ANOVA, F7, 32 = 126.10, P < 0.001). There have been no previous reports of inhibitory effects of Roseobacter spp. on larval settlement. In contrast, Harder et al. (27) found a Roseobacter sp. that induced larval settlement at cell densities in the range of 105 to 106 cells cm−2. However, in a study exploring the use of Phaeobacter sp. strain 2.10 as a probiotic, it was seen that cell extracts from cells at 106 cells cm−2 enhanced scallop larval survival but that extracts from cell densities at 107 cells cm−2 resulted in mortality of Pecten maximus larval cultures, suggesting that this probiotic strain also produces a toxin against the larvae (62).
We found that P. tunicata's inhibition of B. neretina increased with increasing bacterial cell densities of P. tunicata, with the minimum cell density required for inhibition being 1 × 105 cm−2. Dahms et al. (10) identified comparable inhibition of larvae by a Pseudoalteromonas sp. that inhibited settlement of B. neretina at 2 × 105 cells cm−2. Although the possibility of repellents was not discounted by Dahms and coworkers (10), they ascribed the higher settlement of larvae at lower densities of bacterial cells to bacterial free space and lower wettability on biofilm-coated dishes. Similarly, Olivier et al. (54) showed that cyprid settlement in Balanus amphitrite was negatively correlated with total density in biofilms and attributed this to free-space availability. The results described here suggest, however, that antifouling compounds produced by P. tunicata result in inhibition of settlement, as the biofilm established by the WmpR mutant was not inhibitory to larval settlement.
Many studies have stressed the importance of biofilm age in influencing the settlement of larvae or algal propagules (1, 26, 34, 44, 64). However, the rate of settlement reported in these studies seems to correlate with the density of cells, which generally increases with the age of the biofilm. Although changes in cell densities were generally not monitored in these studies, as biofilms grew older, the older the biofilms, the higher the level of induction (1, 26, 34, 44, 64) or inhibition (10, 51). Thus, some invertebrate larvae are able to discriminate between the ages of biofilms as induction or inhibition levels correlate with bacterial cell densities.
A recent study demonstrated that bacterial strains that induce a high level of settlement in the sea urchin Heliocidaris erythrogramma were dominated by the genera Pseudoalteromonas, Shewanella, and Vibrio. These genera were effective at inducing settlement, despite being present at low densities on seaweed surfaces (<1 × 105 cells cm−2) and representing only a small percentage of the total bacterial community (36). Field recruitment of H. erythrogramma correlated with laboratory settlement assays, suggesting that cues were present at a concentration that larvae were able to detect and respond to.
P. tunicata at a range of densities was able to inhibit fungal growth of all three yeast strains (Y1, Y2, and Y3). P. tunicata inhibited the yeast at densities as low as 102 to 103 cells per cm−2 (Fig. (Fig.6),6), suggesting that the antifungal compound is effective at very low concentrations. Y4, a filamentous fungus, was not inhibited by P. tunicata biofilms, and its mycelia extended over the microcolonies. In contrast, the P. tunicata antifungal mutant (FM3) did not affect the attachment of any of the fungal strains in the control biofilm. The results are further supported by the study conducted by Franks et al. (22), who observed that P. tunicata fully inhibited the attachment of Y1 (identified as Rhodosporidium sphaerocarpum) after 24 h of growth in a glass flow cell, whereas the antifungal mutant, FM3, had no inhibitory effect. Furthermore, P. tunicata was able to invade and disrupt an established biofilm of Y1. Purification of the broad-spectrum antifungal component has shown it to be a novel tambjamine molecule (23). In contrast, Phaeobacter sp. strain 2.10 had no antifungal activity at any of the densities tested (102 to 108cells cm−2) (data not shown), and there are no reports of inhibition of fungi by Roseobacter spp. in the literature.
The bacterial attachment assay indicated that P. tunicata biofilms were unable to prevent other marine strains from colonizing onto the surface, regardless of P. tunicata cell density. This was unexpected because in competition experiments conducted in a flow-cell system, P. tunicata was able to outcompete and remove the same strains (58). In glass flow cells (58) and on the surface of the alga (59), competing strains formed separate microcolonies with very little contact between the cells from different species. However, in the bacterial settlement assay, a batch system, the cells formed mixed microcolonies (data not shown).
Biofilms of Phaeobacter sp. strain 2.10 were very effective in preventing the attachment of other bacterial strains, including P. tunicata. Densities as low as 103 cells cm−2 were able to prevent growth of marine bacteria (Fig. (Fig.7).7). This result is consistent with competition studies conducted in glass flow cells (58). Roseobacter strains have been found to display strong in vitro antagonism (4, 29, 61), which is attributed to the production of at least two antibacterial compounds, a peptide (61) and tropodithietic acid (4).
The ubiquity of fouling organisms in the marine environment and the negative consequences of fouling exert strong evolutionary pressures for marine organisms to develop antifouling defenses (68). Alternatively, the acquisition of epiphytic bacteria that produce bioactive compounds against a range of fouling organisms provides the same protection. However, there is little evidence for the efficacy of bioactives produced by bacteria at in situ cell densities against naturally occurring micro- and macrofouling organisms. In this study, cell densities of P. tunicata and Phaeobacter sp. strain 2.10 required to deter settlement and attachment of mico- and macrofouling organisms were determined and it was found that low surface densities of bacteria that produce bioactives are capable of inhibiting fouling organisms. The inhibition of settlement and attachment by these bacteria at low cell densities has not previously been reported.
Skovhus and coworkers (66) showed that P. tunicata is present in very low numbers in the environment, estimating that the absolute abundance of the antifouling clade within the Pseudoalteromonas genus (P. tunicata and P. ulvae) was 1 × 103 cells cm−2. Despite the low density of P. tunicata in the environment, the results presented here confirm that low cell densities P. tunicata are effective at inhibiting settlement of antifouling organisms. Holmström and coworkers (30) demonstrated that 8 of the 10 tested Pseudoalteromonas spp. contained at least one of the four tested antifouling properties: growth inhibition of bacteria or fungi and settlement inhibition of algal spores or invertebrate larvae. It is likely that the antifouling protection of U. australis by bacterial epiphytes consists of several surface-associated Pseudoalteromonas spp. and other inhibitory bacteria, such as Phaeobacter sp. strain 2.10, working as an antifouling consortium, rather than a specific symbiosis with one bacterial epiphyte protecting U. australis. Thus results obtained here suggest that key organisms in an antifouling strategy do not need to be dominant within the epiphytic community in order to have an impact on the colonization of fouling organisms.
Although our data support the contention that natural densities (e.g., 102 to 103) of P. tunicata inhibit colonization of its host's surface, an important caveat is that biofilms established in petri dishes are poor mimics of natural conditions and metabolites may accumulate, in contrast to the field, where they may be dispersed. The importance of relating responses to biologically relevant concentrations has previously been stressed (67), but it is as yet difficult to estimate ecologically relevant concentrations when the microscale localization of these metabolites is not known (15).
It is proposed that the bacteria examined in the present study participate in the antifouling defense, with Phaeobacter sp. strain 2.10 being more effective at inhibiting larvae and bacteria and P. tunicata being more effective against algal spores and fungi. Hence, it appears that U. australis may require both P. tunicata and Phaeobacter sp. strain 2.10 for an effective antifouling strategy. A diverse range of bacteria induce the same settlement response in sea urchin larvae, suggesting redundancy in the function of bacteria on the surface of coralline algae (35-37). Results reported here indicate that although some redundancy in antifouling defense exists within a consortium of Phaeobacter sp. strain 2.10 and P. tunicata, they appear to provide complementary benefits to the host, by targeting different fouling organisms. Thus, a range of epiphytic bacteria that produce bioactives, such as those studied here, can often enhance host fitness.
This research was supported by a USP-AusAID scholarship, the Australian Research Council, The Leverhulme Trust, United Kingdom, and the Centre for Marine Bio-innovation.
We thank A. Franks for fungal strains, S. Egan for antifouling mutant strains, N. Tujula for assistance in the field, and N. Paul for help with statistics.
Published ahead of print on 26 October 2007.