Anthropogenic perturbations in marine microbial communities

Authors


  • Editor: Bernardo González

Correspondence: Balbina Nogales, Microbiologia, Departament de Biologia, Universitat de les Illes Balears, Crtra. Valldemossa km 7.5, 07122 Palma de Mallorca, Spain. Tel.: +34 971 172 068; fax: +34 971 173 184; e-mail: bnogales@uib.es

Abstract

Human activities impact marine ecosystems at a global scale and all levels of complexity of life. Despite their importance as key players in ecosystem processes, the stress caused to microorganisms has been greatly neglected. This fact is aggravated by difficulties in the analysis of microbial communities and their high diversity, making the definition of patterns difficult. In this review, we discuss the effects of nutrient increase, pollution by organic chemicals and heavy metals and the introduction of antibiotics and pathogens into the environment. Microbial communities respond positively to nutrients and chemical pollution by increasing cell numbers. There are also significant changes in community composition, increases in diversity and high temporal variability. These changes, which evidence the modification of the environmental conditions due to anthropogenic stress, usually alter community functionality, although this aspect has not been explored in depth. Altered microbial communities in human-impacted marine environments can in turn have detrimental effects on human health (i.e. spread of pathogens and antibiotic resistance). New threats to marine ecosystems, i.e. related to climate change, could also have an impact on microbial communities. Therefore, an effort dedicated to analyse the microbial compartment in detail should be made when studying the impact of anthropogenic activities on marine ecosystems.

Introduction

Human-derived activities have an important impact on marine ecosystems. The pressures exerted are diverse and result from many activities such as coastal engineering, sediment dredging, pollution, fishing, aquaculture, urban development, maritime transport, tourism, mining, oil extraction, transport and refining of oil, agricultural and industrial activities, etc. (Islam & Tanaka, 2004; Halpern et al., 2007, 2008). All these activities have an effect on all the components of the marine food web, from microorganisms to top animal predators. A recent report has analysed the ecologic impact of anthropogenic activities in the oceans worldwide by focusing on drivers of stress which could be evaluated at a global scale (Halpern et al., 2008). According to that study, more than one-third of the world's oceans (41%) are predicted to be under medium to high impact and certain regions, representing 0.5% of the oceans but a surface of approximately 2.2 million km2, are predicted to be under high impact. Human-derived activities are either directly or indirectly involved in all the threats to ocean ecosystems and in agreement with this, if we consider a global scale, the majority of the areas with lower impact levels have limited human access such as the polar areas (Halpern et al., 2008). All the ocean ecosystems analysed (coral reefs, mangroves, seagrass meadows, seamounts, rocky reefs, soft shallow areas, continental shelf areas, slope areas, pelagic waters and the deep sea) could be considered to be anthropogenically impacted, although to different degrees. The threat scores are higher for the coastal and shelf areas (<200 m depth), which are those more intensively exploited and also receive the impact of land-based activities (Halpern et al., 2008). On a global scale, commercial shipping and fishing-derived activities, together with drivers related to climate change (such as the increase of surface seawater temperature, UV radiation and ocean acidification), have the highest threat scores (Halpern et al., 2008).

Important sources of impact, which were not evaluated in the study of Halpern et al. (2008) because of a lack of global data, include the increase in the frequency and geographic extension of hypoxic zones (areas with dissolved oxygen levels below 2 mL L−1), the effect of coastal engineering, noncargo shipping, aquaculture, recreational fishing and tourism (Cloern, 2001; Halpern et al., 2007; Díaz & Rosenberg, 2008, and supplementary material in Halpern et al., 2008). All these potential sources of impact are also expected to be more important in coastal areas.

Some of the main anthropogenic activities affecting marine ecosystems and their associated pollution risks are shown in Table 1. These pollutants affect both the water column and the sediments, where they concentrate and return to the water after sediment resuspension and by benthic–water flux coupling. There is also a bidirectional exchange with the atmosphere involving the release of compounds to the air and their deposition (Jickells, 2005).

Table 1.   Summary of the risks posed by anthropogenic activities to marine environments
Human activityPollutantsAssociated impact and risks
Agriculture and livestock farmingFertilizers, pesticides, antibiotics, manureNutrient enrichment, eutrophication, hypoxia/anoxia, development of harmful algal blooms, toxicity, bioaccumulation, pathogens, spread of antibiotic resistances
Urban developmentDomestic waste, sewage sludgeOrganic enrichment, eutrophication, pathogens (bacterial and viral), hypoxia/anoxia, development of harmful algal blooms, toxicity by chemicals and heavy metals, loss of amenity and recreational value, litter (solid wastes)
IndustryIndustrial waste, organic pollutants (xenobiotics), heavy metals, radionuclidesToxicity, bioaccumulation, increased atmospheric deposition
Maritime transport and shippingHydrocarbons, xenobiotics (antifouling agents and heavy metals), ballast water, litterToxicity, bioaccumulation, introduction of exotic species, pathogens
Oil extraction and refiningHydrocarbonsToxicity, bioaccumulation
Fossil fuel combustionHydrocarbons, heavy metals, heat, CO2Toxicity, increased atmospheric deposition, warming
Tourism (including recreational nautical activities)Litter (plastic, cigarette buds), faecal waste, nutrients, hydrocarbons, xenobioticsDeath of animal wildlife, accumulation in sediments, hypoxia/anoxia, loss of amenity and recreational value, nutrient enrichment, development of algal blooms, pathogens
Sediment load and dredgingParticles, nutrients, organic pollutants, heavy metalsTurbidity, decrease in light penetration, nutrient enrichment, toxicity, bioaccumulation
AquacultureOrganic load, faecal waste, antibioticsHigh organic load, hypoxia/anoxia in sediments, pathogens, introduction of exotic species, spread of antibiotic resistances, interbreeding of escaped fish
FishingDischarged fishes, pathogens, hydrocarbons, xenobioticsHabitat destruction, decrease of fish stocks pathogens, toxicity, bioaccumulation
Land claiming and deforestation Erosion, increased terrestrial runoff

The effect of perturbations on marine microbial communities is complex, as presented in the conceptual scheme shown in Fig. 1. Most often, multiple stressors (natural and anthropogenic) coexist in a single area, and therefore there is a combination of pollution risks (i.e. eutrophication, hypoxia, chemical pollution, etc.). These stressors have interacting effects that can be additive, synergistic or antagonistic. This provides an idea of the complexity of possible ecosystem responses to perturbations (Crain et al., 2008). On the other hand, it is important to take into account the magnitude and frequency (temporal scale) of the impact, and the fact that in marine ecosystems, the severity of the impact is not always proportional to the amount of the pollutant entering the system. For example, increased nutrient loading does not always result in increased rates of primary production and eutrophication in marine ecosystems. However, this does not mean that the ecosystem is not affected; sometimes subtle effects can be seen such as changes in diversity, in biogeochemical processes and nutrient cycling, in seasonal patterns or in the magnitude and variability of the system (Cloern, 2001; Paerl et al., 2006). The impact of perturbation in marine ecosystems is also dependent on system attributes such as: tidal energy; hydrology (which is controlled by wind, currents, bathymetry, basin geography or river flow and determine seawater residence times in enclosed bays or the spread of a pollutant); optical properties such as water turbidity and the presence of suspended particles; water depth; benthic–pelagic coupling; meteorological and climatic conditions, etc. (Cloern, 2001; Paerl et al., 2006; Cloern & Jassby, 2008).

Figure 1.

 Conceptual scheme of the effect of perturbations on marine microbial communities.

When a microbial community does not change after a disturbance, we can consider it as resistant (Fig. 1). If the community changes in composition due to the perturbation but it recovers (i.e. due to metabolic flexibility and physiological tolerance of microorganisms, rapid growth rates and adaptative evolution) and reverts the original composition, it is considered as resilient (Allison & Martiny, 2008). On the other hand, even if there is a change in community composition, this might not have any effect on its functional performance and on the rates of the different microbial processes. This is explained by the concept of functional redundancy, which implies that different taxa can carry out the same function in the ecosystem and at the same rate. Alternatively, the perturbation can alter the functionality of the community. In this case, we can expect a change in the ecosystem services provided by microbial communities, which usually implies a negative effect on ecosystem health, and on its economic and social value. Allison & Martiny (2008) reviewed the response of microbial communities from different environments to four different types of perturbation in more than a hundred published studies. They showed that in most cases, the disturbance caused a change in community composition, meaning that the communities were not resistant, and that often the communities were not resilient because they did not revert to the composition previous to the impact. There is also evidence showing that changes in microbial composition caused changes in functional processes, which indicates that microbial communities are not always functionally redundant (Allison & Martiny, 2008; Strickland et al., 2009). The structure of the community is also important when facing a situation of environmental stress and it has been demonstrated that communities with high evenness are more likely to be functionally stable after perturbation (Wittebolle et al., 2009). It has also been shown that despite an immediate reduction in evenness in response to a sudden perturbation, bacterial communities tend to revert rapidly to structures of high evenness and richness, even if there has been a change in community composition (Ager et al., 2010).

As mentioned above, the effect of anthropogenic impact on the marine ecosystem is an important issue for maintaining ecosystem services, as well as for habitat sustainability and the development of conservation practices. Therefore, it is currently assessed through a variety of approaches (Foden et al., 2008). The parameters evaluated include physico-chemical (i.e. nutrients, pollutants) and biological (chlorophyll, phytoplankton, zooplankton, benthic invertebrates, algae, seagrass meadows and fishes; Borja et al., 2008). Consequently, there are many reports on the effect of anthropogenic impact on biological components of the marine ecosystem from phytoplankton to fishes, but there is less information on the effects at the level of prokaryotes. Most of the studies assessing the response of microbial communities to perturbation are limited to analysing changes in composition (i.e. community resistance or sensitivity, Fig. 1). Few of them evaluate resilience, i.e. whether or not the community returns to its original composition after the impact. Finally, some studies analyse functional changes, either because they measure process rates or because they analyse the composition of particular functional groups. Few studies address both changes in composition and in functionality. Consequently, there is a wide variety of microbial data available, dealing with a multitude of possible stressors and with different marine ecosystems (i.e. water, sediment), which makes the generalization of patterns difficult. This review discusses several examples of the effect of different human-derived stressors on the composition and functionality of microbial communities in marine environments. The stressors covered in this review are nutrient enrichment, chemical pollution, environmental contamination by antibiotics and the risk of introducing and spreading potential pathogenic microorganisms. The last section discusses risks due to global change (i.e. acidification and temperature increase) or to human activities intended to alleviate global warming, such as ocean fertilization. Throughout the text, except when talking about harmful algal blooms, the terms microbial or microorganism refer only to prokaryotes. Also, the term pollution is used in a broad sense to indicate the multiple possibilities shown in Table 1. However, before discussing the results of some of the studies conducted, it is important to take into account some of the methodological limitations that scientists face when analysing the effect of anthropogenic impact on microbial communities.

Methodological aspects for the assessment of anthropogenic effects at the microbial scale

The analysis of the effect of anthropogenic activities on the composition and activity of microbial communities is complex and ideally requires interdisciplinary expertise. First of all, the study should include the characterization of the abiotic and biotic factors related to the type of pollution occurring at the studied site, particularly those most likely to be influencing microbial activity. Then, the ideal study should have a meaningful design of the sampling procedures (Underwood, 2000). Thus, in order to be able to determine the effects of a particular stressor, data from polluted sites have to be contrasted with those in control areas and therefore comparative studies are imperative. Accordingly, sampling is usually performed along a gradient of pollution and by defining reference stations. However, for a proper design of the sampling strategy, it is necessary to have previous information on the attributes of the area investigated, such as the hydrology of the site (i.e. currents, predominant winds, etc.). For example, information on wind regimes that affect the circulation of water masses in the area of Victoria Harbour (VH) in Hong Kong was helpful in the interpretation of microbiological data (Ho et al., 2008; Zhang et al., 2009). Information on site hydrology is often unavailable and many studies use linear transects from the source of pollution. This strategy has the risk of missing the real plume of the pollutant in the area sampled, as well as the extension of the area impacted, and might lead to erroneous conclusions about the stress caused (Saràet al., 2006). In the absence of hydrological data, an approach that has proven to be useful is the use of techniques from geographic information systems in order to interpret data in a geographical context (Solidoro et al., 2004; Nogales et al., 2007). By defining sampling locations that allow the use of interpolation methods, the variation of abiotic, biotic and microbial parameters can be represented in the form of maps of the area sampled, and this mapping provides more information on the anthropogenic impact in the area than the discrete numerical data. For example, Fig. 2 shows the result of interpolation of simple data (i.e. total number of prokaryotic cells from surface water) on the geography of a coastal area from which no background data were available at the time of sampling. The map clearly highlighted the areas receiving the impact of tourism-dedicated facilities and the stimulatory effect of the effluent of a sewage treatment plant discharging at 30 m depth in the middle of the bay. It is also important to have surveys analysing both the water column and the sediments because there is a tight coupling between these two compartments.

Figure 2.

 Maps showing the geographical variation of total prokaryotic cell counts (DAPI staining) in surface seawater from a coastal location in Mallorca Island (Spain). Interpolation methods were used to represent discrete data obtained following a grid sampling design. Colour legend indicates DAPI counts (× 106 cells mL−1). Geographic coordinates of locations A and B are 39°31′23.40″N, 2°25′22.86″E and 39°30′17.09″N, 2°27′22.66″E, respectively.

Another key aspect to consider is temporal variability. Using interannual data from microbial observatories off the coast of Southern California and the Sargasso Sea, microbial diversity in seawater has been shown to vary temporally according to abiotic and biotic factors with predictable patterns (Morris et al., 2005; Fuhrman et al., 2006). This seasonality has been also reported in coastal and estuarine environments (Kan et al., 2006, 2007; Alonso-Sáez et al., 2007; Nogales et al., 2007; Zhang et al., 2009). However, in near-shore waters receiving anthropogenic influence, there are additional stochastic components of temporal variation due to the irregular occurrence, intensity, variety and duration of stressors (Fig. 1). As a consequence, temporal variation of microbial communities, as well as phytoplankton, in human-impacted environments seems to be more irregular and less predictable in relation to parameters that vary seasonally (Nogales et al., 2007; Cloern & Jassby, 2008; Zhang et al., 2009).

Taking into account the necessity of using molecular methods for analysing microbial community composition, another factor to consider is the level of resolution required. This determines the economic cost, as well as the experimental effort and the time necessary to complete a survey. For comparative studies on microbial diversity, involving samples representing variation at the spatial and temporal scales, the most common approach used so far is the electrophoretic profiling of amplified ribosomal genes or intergenic regions of ribosomal operons (16S rRNA gene and internal transcribed spacer) using primers with specificity at the domain level (i.e. Bacteria or Archaea). These methods provide data on the diversity and structure of microbial communities (Dahllöf, 2002). Besides, the data generated by profiling methods can be analysed numerically using univariate and multivariate statistics (diversity indices, similarity coefficients, cluster analysis, ordination methods, etc.) and can be correlated with environmental parameters, i.e. those indicative of human-derived stress (Ramette, 2007). However, profiling methods are of limited value for identifying the microbial populations present in the samples such as those positively or negatively affected by the stress.

In order to obtain phylogenetic data on the composition of microbial communities, other approaches such as sequencing of cloned 16S rRNA genes or high-throughput sequencing methods should be used (Forney et al., 2004; Cardenas & Tiedje, 2008). In many studies of human-impacted environments, a combination of profiling methods together with cloning and sequencing has been followed. However, the approach of cloning and sequencing of 16S rRNA genes has important drawbacks that limit the conclusions that can be obtained from the data. First of all, the analyses usually involve a limited number of cloned sequences and few samples. Consequently, they fail to cover the huge microbial diversity in the environment (Pedrós-Alió, 2006) and limit the possibility of performing meaningful comparative analyses between samples, both within a particular study (i.e. control vs. polluted at different times) and between studies (i.e. analysis of the same stressor in different locations). In addition, these methods usually overlook the minor components of the communities, which, despite being not abundant, might be functionally relevant or constitute a genetic reservoir that could be important under stress conditions (Bent & Forney, 2008).

On the other hand, 16S rRNA gene sequence data do not allow the inference of physiological and metabolic capabilities of microorganisms, except in the case of particular microbial groups. Therefore, in order to obtain information on community functionality (Fig. 1), some studies try to overcome this limitation using primers or probes specific for certain groups (i.e. ammonia-oxidizing Betaproteobacteria or sulphate-reducing Deltaproteobacteria) with the aim of targeting only those microorganisms involved in metabolic processes with relevance in the area studied. Another strategy is to target functional groups using molecular tools for genes encoding key enzymes (i.e. nifH for nitrogen fixation; nirS/nirK for denitrification; amoA for ammonium oxidation; mcrA for methanogenesis and anaerobic methane oxidation; pmoA for methane oxidation; or dsrAB for sulphate reduction). Irrespective of the target gene used, molecular data alone do not allow one to conclude that a particular metabolism is being carried out in the environment, especially when the molecule targeted is DNA. The most recent and powerful techniques, such as metagenomics, metatranscriptomics and 16S rRNA gene tag sequencing (Cardenas & Tiedje, 2008; Vieites et al., 2009), can help to overcome some of these limitations, but to our knowledge, they have not been applied in studies of human-impacted marine environments so far. However, it is important to keep in mind that for functional assessment of microbial communities, we should combine molecular data with the measurement of environmental process rates (i.e. bacterial production, respiration, pollutant degradation, etc.).

Another important limitation in the interpretation of diversity data from human-impacted environments is our incomplete knowledge of the biogeography of marine microorganisms and the relevant environmental factors determining their distribution. In recent years, there have been some efforts aiming to explain the distribution of marine bacterioplankton and sediment microorganisms (Bowman et al., 2005; Pommier et al., 2007; Fuhrman, 2009). This background knowledge is very important for the interpretation of data from human-impacted environments. A typical result from diversity surveys, including those in polluted environments, is the retrieval of sequence types highly related to sequences from other environments as well as site-specific ones. It is usually impossible to determine whether or not these site-specific sequence types are related to the particular stress found in the site. The direct comparison of these site-specific sequences retrieved in different studies in global, or for a particular ecosystem (i.e. water column, sediment, estuarine, coastal, etc.) or stressor (i.e. nutrient input, chemical pollution, etc.) might help to determine whether or not there are microorganisms that could be systematically associated with human-derived impact. This would allow the definition of indicator microorganisms, as is currently done for higher organisms, to help in the assessment of environmental quality.

Finally, if we really aim to evaluate the effect of anthropogenic perturbations on marine microbial communities at a more global scale (i.e. ecosystem, regional or planetary), it is necessary to understand the microbial response at higher spatial and temporal ranges (Paerl & Steppe, 2003; Li, 2009). So far, we do not know whether the results obtained with the small-scale studies (i.e. with small water volumes of few grams of sediments, taken at relatively small distances and at specific time points) are really representative of larger water bodies or extensive sediment areas, or whether they are representative at long temporal series. To solve this problem, the way forward is the use of remote ocean-observing systems equipped with appropriate sensors, which allow for a continuous, real-time observation of oceanic systems (Paul et al., 2007). For example, satellite imaging can be used to track pollution hazards at the ocean surface over broad spatial scales, i.e. oil spills, sewage effluents, river discharges or algal blooms. For higher resolution, there are moorings and mobile platforms equipped with sensors, which provide data at a smaller spatial scale (Zielinski et al., 2009). Multiple sensors are available, such as CTD (conductivity, temperature, depth), pH, oxygen, redox potential, light, turbidity, chlorophyll and accessory pigments, nitrate, nitrite, ammonium, phosphate, silicate, sulphide, heavy metals (copper, lead, cadmium, zinc, manganese, iron), radionuclides or petroleum hydrocarbons (see Zielinski et al., 2009 for a review). All these parameters are very relevant for the functioning of microbial communities. For example, optical methods for the in situ detection of blooms from the toxin-producing dinoflagellate Karenia brevis have been developed (Paul et al., 2007; Zielinski et al., 2009). Moreover, a field-deployable system called Environmental Sample Processor (ESP) can perform remote molecular analyses of prokaryotic plankton and harmful algal species based in sandwich hybridization assay with a battery of probes with different specificities, and can analyse the presence of the toxin domoic acid in water by a competitive enzyme-linked immunosorbent assay. The device contains also physical and chemical detectors for measurements of basic environmental conditions (Scholin et al., 2009). Another device for performing remote molecular analyses of marine microorganisms is the Autonomous Microbial Genosensor (AMG) (Fries et al., 2007; Paul et al., 2007), which uses an isothermal RNA amplification technique (designated as nucleic-acid sequence-based amplification). It is being tested for the detection of active cells of toxin-producing dinoflagellate K. brevis, based on the amplification of mRNA for the large subunit of ribulose-1,5-bisphosphate carboxylase/oxygenase, rbcL, the key enzyme for carbon fixation (Fries et al., 2007). These platforms, plus further developments that might lie ahead, represent very powerful tools for the in situ and continuously operating analysis of microbial parameters such as changes in composition, functionality or presence of toxins in seawater, which are relevant to analyse anthropogenic impact.

Microbial communities in nutrient-enriched environments

Throughout the text, the term nutrient enrichment is used in a broad sense to refer to the increase of nutrient inputs to the marine system due to human activities (mainly organic and inorganic forms of nitrogen and phosphorus). As shown in Table 1, nutrient enrichment is the result of different practices, such as agriculture, urban development, industries, tourism or aquaculture, and it is one of the most important causes of human-derived impact in the marine environment. To put in perspective the importance of human activities in causing nutrient enrichment, recent estimates show that human activity has increased the rate of formation of biologically available forms of nitrogen by 33–55% (Howarth, 2008). In coastal areas, nutrients are discharged either directly or by river and groundwater flow. They can also reach the marine environment through atmospheric deposition (Jickells, 2005). Besides, due to the movement of water masses, nutrient enrichment can have only local effects or affect distant locations (Saràet al., 2006).

The risk posed by nutrient enrichment is greatest in enclosed bays or seas with limited water exchange, in shallow waters and in estuaries, where differences in water density limit the vertical mixing of the water column (Cloern, 2001). In general, nutrient inputs increase levels of phytoplankton production, stimulate bacterial production and increase oxygen demand and sedimentation rates of particulate material. Two main environmental risks are associated with nutrient enrichment: (1) the development of conditions of hypoxia (<2 mL L−1 dissolved oxygen) or anoxia (no detectable oxygen), either seasonally or permanently in water and sediments (Díaz & Rosenberg, 2008), and (2) the development of harmful algal blooms (Masó & Garcés, 2006). The development of hypoxic or anoxic conditions does not have a detrimental effect on the abundance and activity of bacterioplankton but causes changes in composition and function, as has been demonstrated in waters from the Chesapeake Bay (Crump et al., 2007). Thus, in the absence of oxygen, microorganisms able to use alternative terminal electron acceptors (i.e. nitrate, manganese and iron oxides and sulphate) started to proliferate. At early stages of the development of anoxia, bacterial communities still resembled those in oxic waters, which evidenced that important members of the bacterioplankton in this estuary (SAR11, SAR86 clusters and the picocyanobacterium Synechococcus) have a certain tolerance to anoxia. But, as the most energetically favourable terminal electron acceptors were consumed (i.e. oxygen, nitrate) and sulphate reduction became the predominant respiratory process, there was a dramatic change in bacterial community composition. Phylotypes similar to groups described in sediments, in low-oxygen environments and related to sulphur-oxidizing Gammaproteobacteria were found in the anoxic water (Crump et al., 2007). In sediments, the accentuation of anoxic conditions alters the biogeochemistry and might contribute to increase the flux of inorganic nutrients, i.e. ammonium, silicate and phosphate, from the sediment again to the water column.

Some of the well-studied coastal areas suffering from eutrophication have been reviewed in recent publications: the Baltic Sea, the northern Gulf of Mexico, the East China Sea, the Northern Adriatic Sea, the Chesapeake Bay and the Neuse River estuary (Paerl, 2006; Paerl et al., 2006; Rabalais et al., 2010). Interventions to reduce diffuse nutrient loads in coastal areas have been carried out but they do not seem to have an immediate effect on the ecosystem, although they have alleviated the problems in some cases (Ho et al., 2008; Rabalais et al., 2010). In the following subsections, we discuss some examples of the effect of nutrient enrichment due to human activity, where the experimental design allowed the comparison of samples subjected to different levels of impact.

Microbial communities in areas receiving both point and diffuse land-related nutrient inputs

Several comparative analyses of microbial diversity along pollution gradients in water and sediments of coastal areas receiving high nutrient inputs have been published in the last years (Paerl et al., 2003; Kan et al., 2006, 2007; Urakawa et al., 2006; Zhang et al., 2007, 2009). A common characteristic to many of these environments is the presence of point sources of pollution (i.e. discharges from sewage treatment plants, industries, etc.) as well as diffuse sources of nutrients such as river discharges, runoff from land, harbour activity, atmospheric deposition, etc., hence their complexity. Often, these are transitional environments where we find strong gradients in important variables such as salinity (ranging from that typical of freshwater to those characteristically marine), as for example in estuaries whose high productivity is in part due to the nutrient load received from river discharge. Because of the natural heterogeneity of estuaries, it is difficult to relate the observed changes in microbial community only to nutrient enrichment in these environments. This complexity is well exemplified by the impossibility of finding common trends in the composition of microbial communities in estuaries studied worldwide, as discussed by Kan et al. (2007). These authors interpreted that the high spatial and temporal variability of communities in estuaries probably reflect strong regional or physiographic components. Therefore, in order to simplify the number and complexity of variables involved in determining the composition of microbial communities in nutrient-enriched environments, only examples of environments where truly marine samples were analysed along gradients of nutrients will be presented.

A particularly interesting study has been conducted in VH in Hong Kong because it has allowed the comparison of environmental data before and after the implementation of sewage treatment to water discharged to the harbour. The area is also a good example of coastal development and exploitation; it is located next to the big city of Hong Kong and includes one of the busiest ports in the world. Up to November 2001, it received untreated local sewage effluents discharged in VH [over 1.5 million tonnes of sewage per day (Zhang et al., 2007)]. From that date, 70% of the sewage is subjected to treatment (53% chemically enhanced primary treatment and 17% with secondary treatment) and the discharge site has been moved to the west of the harbour (Ho et al., 2008). In addition, this area receives high nutrient loads (i.e. nitrate and silicate) from Pearl River, particularly during the rainy season (summer). The predominant wind conditions in summer push the river plume towards VH and also cause upwelling of deep coastal water, intensifying eutrophication. The situation is the opposite in winter, when rainfall is low, predominant winds push the river plume away from the area of VH and cause downwelling of surface waters (Ho et al., 2008). The discharge of preliminary-treated sewage water caused an increase in ammonium and phosphate concentrations in the area around VH throughout the year (7–20 μM ammonium and 0.7–1.4 μM phosphate). However, these values represented an average decrease of 2–10 μM of ammonium and 0.1–1.0 μM of phosphate in the waters of VH with respect to the period of sewage discharge without treatment (Ho et al., 2008). There was also an increase in the concentration of dissolved oxygen in deep waters, which indicated better water quality in the area of VH. The impact of nutrient load was also evident in the sediments. Thus, organic matter content, exchangeable ammonia (NH3-N) and available phosphorus were also higher in sediments of the harbour samples (Zhang et al., 2008b). These sediments also contained higher amounts of other pollutants such as polyaromatic hydrocarbons (PAHs) and heavy metals (Zhang et al., 2008a, b), which could be directly related to shipping activities in VH. Redox potential was more negative in VH sediments as we could expect from nutrient-enriched environments. This, together with the higher amounts of acid volatile sulphide and total sulphur measured in VH sediments, indicated higher sulphate reduction rates in the polluted sediments.

Microbial communities in seawater and sediments of VH area were analysed in the same period (2004–2006), about 2.5 years after sewage treatment started, in transects from the West (PC, the area most influenced by Pearl River) to the East coast (TLC site) and the South China Sea, including three locations in the harbour. In agreement with the higher nutrient concentration at VH, the number of prokaryotic cells in seawater and sediments was higher. Bacterial communities in VH seawater were different from those at the neighbouring sites and the same was true for the sediments (Zhang et al., 2007, 2008a, b). Seasonal variation in the bacterioplankton composition was observed in these samples, mainly driven by changes in temperature, with the exception of the Western part of VH (site VHW), which did not show the typical seasonal trend. At site VHW, which is the one receiving the impact of the sewage effluents more directly, the main drivers of bacterioplankton dynamics were factors indicative of impact such as the amount of dissolved oxygen, phosphate and organic content (Zhang et al., 2009). Parameters indicative of pollution were also important in determining the composition of bacterial communities in sediments and they influenced temporal variability as well (Thiyagarajan et al., 2010). Detailed phylogenetic analysis of 16S rRNA gene clone libraries in seawater identified sequences related to potential human pathogens, such as Arcobacter or with likely faecal origin (Bacteroides) in the harbour (Zhang et al., 2007). Other distinctive characteristics of the bacterioplankton in the harbour were the dominance of sequences of Gammaproteobacteria (with a high proportion of Oceanospirillales, Alteromonadales, Enterobacteriales and Vibrionales) over those of Alphaproteobacteria, which is usually the main group in seawater, and the lack of sequences of typical coastal marine microorganisms such as the SAR11 group (Zhang et al., 2007, 2009). This assemblage differs remarkably from those described in coastal environments (Pommier et al., 2007), and agrees with the nutrient-enriched characteristics of this area. Significantly different microbial assemblages also developed in the most polluted sediments and there were also differences in the composition within the functional group of the sulphate-reducing bacteria (SRB) (Zhang et al., 2008a, b). Diversity was high, both in water and sediment of VH, and therefore, pollution did not seem to cause a decrease in the overall diversity, but the contrary. In fact, high diversity and high equitability seem to be a common feature in water and sediments of other human-impacted environments (Nogales et al., 2007; Borin et al., 2009), probably because perturbation generates conditions for the proliferation of several different microorganisms.

Recent studies conducted in the eutrophicated Venice lagoon also highlight areas with different trophic status, some of them under the strong influence of tributaries and point discharges of urban and industrial effluents (Solidoro et al., 2004). These studies also demonstrate differences in the microbial community composition under different levels of anthropogenic impact (Borin et al., 2009; Celussi et al., 2009), although the differences seem to be evident only in the sediments. The main driver affecting community composition in sediments seemed to be, as in VH, the discharge of urban waste, which in this case favoured the development of Vibrionaceae (Borin et al., 2009). In contrast, differences in microbial composition along the pollution gradient were not evident in the water column because the temporal variation in bacterioplankton composition in seawater was higher than that observed at the spatial scale when comparing assemblages from areas with different levels of pollution (Celussi et al., 2009). This lack of spatial resolution might be due to the particular hydrologic characteristics of the Venice lagoon, which is highly dynamic, or to the methods used to explore diversity. What seems to be true is that the lagoon holds particular bacterioplankton communities, different from those in incoming water from the North Adriatic Sea. These communities, as in the case of VH, are enriched in copiotrophic microorganisms such as gammaproteobacterial genera Vibrio, Alteromonas and Pseudoalteromonas as well as members of the Bacteroidetes (Simonato et al., 2010). Interestingly, bacterial activity (measured as bacterial production and hydrolytic ectoenzymatic activities) inside the lagoon appears to be relatively constant despite seasonal differences in community composition and water parameters (Celussi et al., 2009), which would agree with the hypothesis of functional redundancy of these microbial communities (see the scheme in Fig. 1).

The examples just reported, together with estuaries, are complex and variable ecosystems with many factors besides nutrient enrichment interacting and having an effect on microbial communities and high nutrient loads. However, some of the trends observed in these environments, in particular in relation to microbial composition, also hold for simpler environments. An example of those is a small oligotrophic bay, Cala Penyes Rotges (Mallorca Island, Spain), which is exploited for tourist purposes during the summer season; it receives the impact of housing development of the neighbouring coast, the use of a small beach and of a nearby recreational marina, but no discharges of sewage treatment plants, agriculture, aquaculture or river discharges (Nogales et al., 2007; Aguiló-Ferretjans et al., 2008). In this case, nutrient loads are low compared with the environments discussed above and the ecosystem is far from being eutrophicated. However, as in the more complex environments (Zhang et al., 2007; Simonato et al., 2010), bacterial communities in surface seawater were shown to be highly diverse and dynamic on the temporal scale, and less predictable by changes in important seasonal parameters, such as temperature. At times when perturbation was higher (i.e. summer), the composition of bacterial communities deviated clearly from that of the reference sites: a high proportion of sequences from the Gammaproteobacteria and Bacteroidetes, as well as a reduction of Alphaproteobacteria, was observed; typical oligotrophic groups such as SAR11 or SAR86 decreased or disappeared, and the main alphaproteobacterial species belonged to the Roseobacter clade (Nogales et al., 2007; Aguiló-Ferretjans et al., 2008). These patterns are the same as those found in VH and the Venice lagoon (Zhang et al., 2007; Simonato et al., 2010), even though the amount of nutrients entering this oligotrophic bay is much lower. Therefore, independently of the particular phylotypes observed in each site (which appeared to be different after sequence comparison), there seems to be a general response of microbial communities to nutrient enrichment, at least in seawater, which seems to be independent of the absolute magnitude of the nutrient load. Thus, there seems to be a counter-selection against typical oligotrophic microorganisms and the proliferation of copiotrophic bacteria able to grow and respond to increased nutrient concentrations, such as certain groups of Gammaproteobacteria (i.e. Alteromonadales, Pseudoalteromonas, Pseudomonadales, OM60 group), members of the division Bacteroidetes and representatives of the Roseobacter clade in the Alphaproteobacteria. All these groups have been associated with experimental nutrient enrichment or algal blooms in marine environments (Eilers et al., 2000; Pinhassi et al., 2004; Buchan et al., 2005). These microorganisms probably react directly to nutrient enrichment by increasing their growth rate and their biomass, following the typical strategy of fast growers. However, they can also proliferate in response to changes in abundance and/or composition of phytoplankton, and probably, there is a tight coupling between the proliferation of certain types of algae and particular microbial populations, as there is a coupling between phytoplankton primary production and bacterial production. None of the studies conducted so far have explored at the same time the diversity of phytoplankton and bacterioplankton in nutrient-enriched environments, and, therefore, this aspect remains unknown.

Impact of point sources of nutrients on marine microbial communities: the example of aquaculture

A good model for studying more directly the effect of nutrient enrichment in marine ecosystems without the presence of additional stressors is aquaculture. Fish production in aquaculture farms is a growing economical sector worldwide but has the drawback of releasing high amounts of organic matter (resulting from uneaten fish food and faecal material) to the seawater and the sediments below the cages. Because aquaculture has been shown to have an effect on environmental quality and health, there are numerous studies addressing this issue.

Meta-analysis of published ecological data from the water column, as well as studies conducted in particular locations, have shown that fish farms have a significant local effect on the pools of particulate and dissolved organic and inorganic nutrients, such as particulate organic phosphorus, particulate organic nitrogen, dissolved organic nitrogen, dissolved organic carbon, ammonium and nitrite (La Rosa et al., 2002; Sarà, 2007a, b; Garren et al., 2008; Navarro et al., 2008). Usually, no effects are observed on the concentration of phosphate and silicate. As in most complex environments nutrient enrichment from fish cages causes a significant increase in bacterioplankton abundance and heterotrophic production (Sakami et al., 2003; Sarà, 2007b; Garren et al., 2008; Navarro et al., 2008) as well as in the abundance of virus-like particles (Garren et al., 2008). Although increases in phytoplankton abundance is not always evident from chlorophyll data, a stimulatory effect of fish cages on autotrophic eukaryotic phytoplankton has been observed as well, although this does not lead to the development of phytoplankton blooms because there seem to be a tight control by grazing (Navarro et al., 2008; Pitta et al., 2009). Hydrology is probably also playing a role because fish farms are usually located in areas with good water exchange.

Despite all these important perturbations of the marine microbial food web, few studies have reported which are the changes in microbial communities that occur under fish cages in comparison with control areas. In a study performed in farms of milkfish (Chanos chanos) in the Philippines, Garren et al. (2008) demonstrated that the composition of bacterial assemblages (both free-living and particle-attached fractions) in fish pens was different from that of seawater from different locations in their proximity. There were few phylogenetic groups represented in libraries from the fish pens and the proportion of clones affiliated to the Cyanobacteria, the Proteobacteria and the Bacteroidetes were high. The results of this study would indicate a reduction in the bacterioplankton diversity near fish cages in comparison with surrounding areas, contradicting the findings reported above about microbial diversity in nutrient-enriched coastal areas. This apparent reduction in bacterial diversity due to aquaculture should be confirmed in other systems because it might not be a general finding. For example, Wei et al. (2009) observed high bacterial diversity in fish ponds dedicated to the culture of two different species (grouper, Epinephelus diacanthus; and abalone, Haliotis diversicolor supertexta), and they related this to the composition of the feeding supplied. However, it should be taken into account that fish pens and ponds are completely different aquaculture systems and therefore the effect on microbial communities might also be different. On the other hand, if we take into account the fact that the perturbation caused by a fish farm is more constant in amount, composition and periodicity of additions (i.e. fish food and fish faeces), we can hypothesize that this might lead to the development of microbial communities highly specialized in processing this particular type of organic load, maybe dominated by a few very efficient microorganisms, and therefore less diverse.

The most dramatic effect of the nutrient load from fish farms occurs in sediments below the cages. There, the high organic load, mainly in the form of particulate organic matter, increases oxygen demand and alters the biogeochemistry of the sediments. The granulometry of the sediments is crucial for determining the severity of the impact (Kalantzi & Karakassis, 2006) and the hydrology of the site is important for determining its spatial dispersion (Saràet al., 2004, 2006). The benthic biomass in environments affected by fish farms becomes dominated by microbial components, and bacterial abundance increases significantly compared with control areas (Mirto et al., 2000; La Rosa et al., 2001; Bissett et al., 2007; Castine et al., 2009) and decreases again with decreasing organic loads such as during fallowing (Bissett et al., 2007). In contrast, fish farms cause a decline of some groups of benthic fauna and seagrasses (Mirto et al., 2000; La Rosa et al., 2001; Holmer et al., 2008).

With respect to functionality, the organic load of fish farming stimulates microbial anaerobic respiration processes in sediments below fish farms, such as sulphate, iron and manganese reduction, denitrification, methanogenesis and fermentation (Christensen et al., 2000; Holmer et al., 2003; Bissett et al., 2009). By far the most important anaerobic process in fish-farm sediments, as in other nutrient-enriched environments, is sulphate reduction because sulphate is usually present in nonlimiting concentrations in marine sediments (Holmer et al., 2003; Asami et al., 2005). The production of high amounts of hydrogen sulphide by sulphate reduction, concomitant with a decrease in oxygen, might lead to the inhibition of important microbial processes in the nitrogen cycle such as nitrification (oxidation of ammonium to nitrate), and hence, to lower the rates of coupled denitrification (reduction of nitrate to nitrogen gas), which requires oxidized forms of nitrogen. Coupled nitrification–denitrification is important for reducing the high nitrogen load of fish-farm sediments and for controlling the efflux of ammonium from the sediments to the water column (McCaig et al., 1999; Christensen et al., 2000). However, in reduced fish-farm sediments, the process of dissimilatory nitrate reduction to ammonium, which also needs oxidized forms of nitrogen, can be quantitatively more important than denitrification. As a result of reduced nitrification, there is less nitrogen removed from the sediment as nitrogen gas (ultimately released to the atmosphere) and a higher efflux of ammonium to the water column, which stimulates primary production and maintains the nitrogen in the system (Christensen et al., 2000). But even if nitrification is not inhibited, it might be insufficient to oxidize all the ammonium resulting from the organic load to nitrate, and consequently there is a net flux of ammonium to the water column (Bissett et al., 2009).

The bacterial composition in sediments under fish farms along gradients of organic pollution, and hence sulphide and oxygen concentrations, have been analysed by 16S rRNA gene-based approaches in Japan (Asami et al., 2005; Kawahara et al., 2009) and Australia (Bissett et al., 2006; Castine et al., 2009). Significant differences in composition of clone libraries from fish-farm and reference sediments are usually observed, both at a broad and at a fine phylogenetic resolution (Bissett et al., 2006, 2007). These studies have also shown that bacterial community composition changes significantly after cessation of production, such as in periods of fallowing, but the communities do not revert to the composition before the impact, i.e. they are not resilient (Bissett et al., 2006, 2007; Bissett et al., 2008). A distinctive characteristic composition of bacterial communities in fish-farm sediments cannot be easily drawn due to the high microbial diversity found in marine sediments (Asami et al., 2005; Bissett et al., 2006; Castine et al., 2009; Kawahara et al., 2009). However, even when these farm sediments were located in different areas and they were dedicated to the production of different fish species, there were some common findings that allow the formulation of certain general trends. For example, an increase in the abundance of Deltaproteobacteria (which includes most of the sulphate-reducing genera) is generally observed. This has been proven by a higher proportion of deltaproteobacterial 16S rRNA gene clones in fish-farm libraries (Castine et al., 2009; Kawahara et al., 2009) or by quantification of SRB using the α-subunit dissimilatory sulphite reductase (dsrA) gene as target (Kawahara et al., 2008). Besides, in agreement with a high production of sulphide by sulphate reduction, a characteristically high proportion of clone sequences of putative sulphur-oxidizing Gammaproteobacteria (Asami et al., 2005), and the formation of mats of the sulphur-oxidizer bacterium Beggiatoa on the surface of fish-farm sediments, is often reported (Karakassis et al., 2002; Bissett et al., 2006, 2007). Sequences of the Epsilonproteobacteria are also abundant (Bissett et al., 2006; Castine et al., 2009; Kawahara et al., 2009). This group is also related to the metabolism of reduced sulphur compounds and it is usually found in sulphide-rich sediments (Campbell et al., 2006). The proportion of clones affiliated to the division Bacteroidetes (Flavobacteria in particular), which are supposed to play a role in biopolymer degradation, tends to be high in fish-farm sediments as well, and the phylotypes recovered are different from those in control areas (Asami et al., 2005; Bissett et al., 2006; Bissett et al., 2008; Kawahara et al., 2009). Finally, clone sequences related to betaproteobacterial nitrifiers are usually not recovered in libraries from fish-farm sediments targeting total bacterial diversity (Bissett et al., 2006; Kawahara et al., 2009), in agreement with the low rates of ammonia oxidation measured in sediments (McCaig et al., 1999) and their low relative abundance in the community (Bissett et al., 2006). However, using molecular tools targeting selectively this bacterial group, several authors have demonstrated a different composition of the betaproteobacterial nitrifiers in fish-farm sediments compared with control sites (McCaig et al., 1999; Bissett et al., 2009). These changes in the composition of microorganisms participating in the degradation of the organic load and in the sulphur and nitrogen cycles, together with the changes in process rates (i.e. sulphate reduction), provide good evidence of an altered functionality of the microbial communities in sediments below fish farms. The perturbation is likely to have an effect on archaeal groups too, but the composition of archaeal communities or their role in biogeochemical processes such as ammonia oxidation or methanogenesis in these environments has not been explored so far.

Taking together the results of the studies conducted in environments receiving point and diffuse nutrient inputs, some general effects of nutrient enrichment on microbial community composition and function can be proposed (Fig. 3). Thus, despite the composition of the microbial assemblage developing at a particular site, it seems clear that the structure of the microbial food web will be affected as well as important biogeochemical cycles, such as the nitrogen, carbon and sulphur cycles.

Figure 3.

 General effects of two examples of human-derived stressors on marine microbial communities.

Impact of pollution with organic chemicals, hydrocarbons and heavy metals

Persistent organic pollutants, such as herbicides (aldrin, DDT, hexachlorobenzene, etc.), polychlorinated biphenyls (PCBs), dioxins, dibenzofurans and hexachlorocyclohexane, as well as heavy metals and petroleum hydrocarbons, contaminate the water column and sediments worldwide (Lohmann et al., 2007). They reach the marine environment by direct discharges (from coastal areas or dumping into the sea), runoff from land, river discharges or atmospheric deposition. The pollutants accumulate in sediments and in the sea-surface microlayer at the seawater–atmosphere interface (Wurl & Obbard, 2004). They also adsorb to floating and stranded plastics, which contributes to their transport and persistence in the environment (Rios et al., 2007; Teuten et al., 2009).

Hydrocarbon pollution

Among the different pollutants entering the marine environment, the effect of hydrocarbon pollution on marine microbial communities has been the subject of numerous studies, mainly driven by the concerns caused by tanker accidents such as that of the Exxon Valdez in Alaska or the more recent accident of the Prestige tanker in Spain. Most of these studies have focused on analysing the effect of crude oil spills on microbial diversity, the changes in response to bioremediation trials (usually involving the addition of nutrients) and the extent of hydrocarbon biodegradation in polluted environments. These topics are covered in excellent recent review articles (Head et al., 2006; Yakimov et al., 2007) and will not be treated in detail here. The results obtained show that the microbial communities developing in response to events of hydrocarbon contamination differ in composition although the efficacy in hydrocarbon removal is similar (Head et al., 2006). It is also characteristic that the communities do not converge to the composition that they had previous to the pollution, and therefore again they are not resilient (Allison & Martiny, 2008).

Events of acute contamination cause a reduction of microbial diversity in the short term. This is due to two main reasons: the disappearance of certain groups of microorganisms (i.e. archaea and cyanobacteria) and the strong selection for specialist hydrocarbon-degrading marine bacteria (i.e. Alcanivorax, Cycloclasticus), which become predominant particularly when nutrients are added to stimulate hydrocarbon degradation (Head et al., 2006; Yakimov et al., 2007). In contrast, microbial diversity is usually high in chronically or long-term hydrocarbon-polluted marine environments (Hernandez-Raquet et al., 2006; Nogales et al., 2007; Zhang et al., 2007; Paisséet al., 2008; Alonso-Gutiérrez et al., 2009), as expected for communities adapted to the presence of hydrocarbons in the environment. This behaviour is exactly the same as that observed in environments polluted chronically with nutrients (Zhang et al., 2007), as explained in the previous section, and seems to be an intrinsic property of microbial communities in chronically perturbed environments.

Typically, 16S rRNA gene sequences related to known hydrocarbon degraders [i.e. Alcanivorax, Cycloclasticus, etc. (Yakimov et al., 2007)] are not (or rarely) detected in chronically hydrocarbon-polluted environments. This might indicate that those hydrocarbon degraders are minor components of these communities, and therefore are missed because of method limitation, although they become predominant in bioremediation trials. For example, sequences affiliated to Alcanivorax were not observed in mesocosms prepared with seawater from Messina Harbour but they became predominant 15 days after the addition of oil and nutrients (Cappello et al., 2007). Also, most-probable-number counts of hydrocarbon degraders in two marinas in the United States were shown to be low (Piehler et al., 2002). Recent studies conducted in areas polluted after the Prestige tanker oil spill sampled one year after the accident revealed the importance of members of the suborder Corynebacterineae (i.e. Rhodococcus) and the family Sphingomonadaceae in the degradation of the alkane and aromatic fraction of this heavy oil, respectively (Alonso-Gutiérrez et al., 2009). These results show the wide diversity of bacterial hydrocarbon degraders in environmental samples and the importance of certain groups (i.e. Actinobacteria) as hydrocarbon degraders in rough environments (i.e. rocks polluted with heavy oil). Alternatively, there might be novel, uncharacterized degraders in these polluted environments (Paisséet al., 2008). We have experimental evidence supporting this hypothesis coming from an analysis of aromatic ring-hydroxylating dioxygenase genes (ARHD) in polluted coastal sediments from Patagonia in Argentina (Lozada et al., 2008). Sequences representative of the already known phnAc-like genes of Alcaligenes faecalis AFK2, phnAI-like genes of Cycloclasticus spp. and nahAc-like genes of Pseudomonas spp., were detected in these sediments. In addition, five novel types of putative ARHDs with 58–65% sequence similarity to known ARHDs were obtained. These novel ARHDs contained the conserved residues of bacterial ring-hydroxylating dioxygenase α-subunits (Lozada et al., 2008) and therefore they might represent enzymes with novel catalytic properties or substrate specificity that might be hosted in as yet unrecognized bacterial genera.

We have a biased perception of the risk associated to hydrocarbon pollution towards contamination caused by tanker accidents and this has motivated most of the research on hydrocarbon pollution and bioremediation studies conducted so far. However, the contribution of tanker accidents to total hydrocarbon contamination in the sea is very low. For example, in the Mediterranean Sea, 80 000 tonnes of oil were spilled in accidents between 1990 and 2005 [Environmental European Agency (EEA), 2006]. In contrast, 250 000 tonnes are discharged per year due to shipping operations (i.e. discharges of ballast water, tank washing, oil sludge, bilge water or engine room wastes) and about 120 000 tonnes per year from oil terminals and routine land-based operations (EEA, 2006). Therefore, it is important to take into account this type of pollution and to assess properly its effects on coastal microbial communities.

An example of a study of the effect of land-based sources of hydrocarbon pollution on microbial communities is the one done in a chronically polluted coastal retention basin in the Mediterranean Sea (Etang de Berre, France). This coastal lagoon receives hydrocarbons from refineries, petrochemical plants and transportation systems. Samples were taken along the hydrocarbon pollution gradient (Hernandez-Raquet et al., 2006; Paisséet al., 2008). As in environments where pollution was due to nutrient enrichment, bacterial communities in sediments of the lagoon changed in relation to the hydrocarbon pollution gradient, and about 32% of the variation could be explained by hydrocarbon concentrations in water and sediment (Paisséet al., 2008). This means that, although other environmental factors were determining microbial composition in the area, hydrocarbons appeared as a key factor for diversification of communities. When the composition of bacterial communities was analysed in detail, a predominance of sequences related to Delta- and Gammaproteobacteria, as usually observed in marine sediments, was obtained, although variability in different sampling years was high as happens also in nutrient-enriched environments (Hernandez-Raquet et al., 2006; Paisséet al., 2008). As usual, only a few sequences could be related to hydrocarbon-degrading bacteria and these included sequences related to Marinobacter spp., a spirochaete, and oil-degrading SRB (Paisséet al., 2008). Sequences related to putative hydrocarbon-degrading SRB have also been retrieved in polluted sediments from several coastal locations including the United States, Italy, Venezuela, Puerto Rico and South Korea and in harbour sediments (Pérez-Jiménez & Kerkhof, 2005; Chin et al., 2008). Therefore, anaerobic metabolism of hydrocarbons by SRB seems to be common in polluted environments.

As mentioned above, activities derived from shipping are important sources of hydrocarbon contamination in the marine environment (EEA, 2006; Halpern et al., 2008). A good proportion of the pollution caused by ships is due to illegal discharges. In the Mediterranean, for example, the density of spills can be correlated with major shipping routes (Ferraro et al., 2009). The effect of this type of diffuse hydrocarbon pollution in the composition of seawater microbial communities has not been addressed but it might be a factor to consider for explaining the described cosmopolitan distribution of specialist hydrocarbon degraders in the marine environment (Yakimov et al., 2007). In areas exploited for tourism, there is a consistent increase in the number of hydrocarbon spill detections during the high season, correlating with the increase in boating activity (Ferraro et al., 2009).

The pollution caused by maritime transport and recreational boats is likely to be a source of fuels (i.e. diesel) and synthetic lubricants to seawater and sediments. There are few studies addressing the effect of this type of pollution. One experiment was performed in in situ mesocosms with water and sediments from a tropical estuary in Singapore, impacted by boating-derived activities. The mesocosms were treated with diesel oil at concentrations reproducing the mean and the highest hydrocarbon concentrations measured in the area (Nayar et al., 2005). The response observed was negative for eukaryotic phytoplankton and picocyanobacteria (Synechococcus) at higher hydrocarbon concentrations, although picocyanobacteria were stimulated at the lower concentrations. In contrast, the number of heterotrophic bacteria and production rates increased in response to the treatment, particularly at the higher concentrations (Nayar et al., 2005). These results show that there is a rapid response of microbial communities to diesel pollution in the marine environment, as demonstrated for more complex hydrocarbon mixtures such as crude oil (Head et al., 2006). A second study tested the short-term effect of diesel oil, a biodegradable lubricant and a synthetic lubricant (clean and used) in a field experiment in pristine Antarctic sediments (Powell et al., 2005). Although the bacterial communities differed from the control in all treatments, the most significant variations were due to treatment with diesel, followed by synthetic lubricants, but not with the biodegradable lubricant (Powell et al., 2005). These results show ways of reducing the impact of lubricants in the marine environment, particularly in sensitive areas such as the poles.

Maritime activities require the building of ports and recreational marinas. Because of the constant increase of maritime transport (United Nations Conference on Trade and Development, 2009) and the interest in increasing tourism-derived activities (European Commission, 2009), there is a constant pressure for the development of harbours and recreational marinas, and therefore, we can expect an increase in the environmental stress posed by them. Harbours are complex and chronically polluted habitats and constitute a source of pollution for the surrounding environment (Commendatore & Esteves, 2007). Some of the pollutants that are commonly found in harbours are hydrocarbons (i.e. as result of boat traffic, accidental spills or discharge of bilge oil and ballast water), detergents, surfactants or antifouling compounds (i.e. heavy metals, biocides), even though ports have facilities for the collection of wastes and for pollution control. Additionally, they receive nutrient inputs, either directly due to harbour activities or through discharges of rivers and sewage effluents such as VH in Hong Kong (Zhang et al., 2007; Ho et al., 2008). Harbour seawater and sediments have been shown to hold particular bacterial communities, different from those in adjacent areas, highly diverse and highly variable at the temporal scale (Schauer et al., 2000; Denaro et al., 2005; Nogales et al., 2007; Zhang et al., 2007; Aguiló-Ferretjans et al., 2008; Zhang et al., 2008a, b; Ma et al., 2009). Despite chemical pollution, the effect of nutrient enrichment seems to be the most important factor driving the composition of bacterioplankton in harbours, and for example nutrient enrichment might explain the high abundance of sequences related to Clostridium and Vibrio in the sediments of Milazzo Harbour, subjected to high organic load (Yakimov et al., 2005). Relationships between bacterioplankton community composition and hydrocarbon pollution were only found in the study conducted in Messina Harbour, where the relative abundance of putative hydrocarbon-degrading bacteria correlated with an increase in hydrocarbons (Denaro et al., 2005). In sediments, stimulation of hydrocarbon degraders has been observed after hydrocarbon addition, indicating that there is a primed microbial community able to use these pollutants, both aerobically and anaerobically (Hayes et al., 1999; Hayes & Lovley, 2002; Yakimov et al., 2005). As in other hydrocarbon-polluted sediments, SRB seem to be involved in hydrocarbon degradation in harbour sediments. For example, a recent study on the composition of the active sulphate-reducing assemblages in sediments of Boston Harbour by analysing transcripts (mRNAs) of dsrAB genes (Chin et al., 2008) found sequences closely related to cultured SRB capable of metabolizing aromatic compounds such as benzoate or naphthalene.

Heavy-metal pollution

Several studies have addressed the composition of bacterial communities in marine sediments polluted with heavy metals. As in the case of other pollutants, bacterial communities in sediments with different levels of heavy-metal pollution are shown to differ (Gillan et al., 2005; Morán et al., 2008; Toes et al., 2008). The same result has been observed for archaeal and photosynthetic communities (Toes et al., 2008), as well as for bacterial assemblages in seawater, rocks and associated with the algae Ulva compressa in coastal sites from Chile contaminated with copper (Morán et al., 2008). Different from the cases of nutrient or hydrocarbon pollution, which seem to stimulate bacterial growth, total prokaryotic cell numbers appear to be negatively correlated with heavy metals such as cadmium, copper, zinc and lead (Gillan et al., 2005; Gillan & Pernet, 2007). However, when the calculations were done for particular phylogenetic groups, few statistically significant correlations could be established, and they seemed to be dependent on the sample. Therefore, it is difficult to determine the effect of heavy metal on particular groups of microorganisms with the data available so far. In highly polluted sediments of a Norwegian fjord, the abundance of Gammaproteobacteria (determined by FISH counts) was negatively correlated with copper and zinc, and that of Bacteroidetes with copper, zinc and cadmium. In contrast, in Belgian sediments with lower pollution levels, the only significant correlation was found between Bacteroidetes and cadmium (Gillan et al., 2005; Gillan & Pernet, 2007). In these two sediments, bacterial community composition was also analysed by a cloning and sequencing approach. The results evidenced the presence of the main groups typically found in marine sediments: Gamma- and Deltaproteobacteria and Bacteroidetes. But some of the clones in the Norwegian sediment grouped with clones from an Antarctic sediment polluted with heavy metals and hydrocarbons, and a group of gammaproteobacterial sequences was found in both the Norwegian and the Belgian sediments (Gillan et al., 2005; Gillan & Pernet, 2007). This might indicate that certain bacterial populations might be favoured in heavy-metal-polluted sediments, although these relationships are even more difficult to draw than in the case of hydrocarbon pollution.

A laboratory experiment was performed in microcosms to simulate the effect of sediment disturbance (dredging) in a sediment chronically polluted with heavy metals. In addition, the effect of deposition of a 3 mm layer of polluted sediment on the surface of unpolluted sandy sediments was also analysed (Toes et al., 2008). These two practices (dredging and deposition onto a different location) are common in the marine environment. Homogenization of the polluted sediment (simulating dredging) caused minor changes in the composition of microbial communities. However, overlying of sandy sediments with polluted sediment caused significant changes in the composition in the three microbial components analysed (Bacteria, Archaea and Cyanobacteria) in comparison with the unpolluted control, although the effect was lower for Archaea (Toes et al., 2008). After one year of incubation, some common bacterial groups were observed in clone libraries from the artificially polluted sandy sediments in comparison with those from the chronically polluted sediment: representatives of the Roseobacter clade, the genus Vibrio and a member of the Flavobacteriaceae whose sequence was related to a clone from a heavy-metal-polluted Antarctic sediment (Toes et al., 2008). These sequences might represent microorganisms favoured by the conditions imposed by heavy-metal pollution, although to confirm this, it will be necessary to corroborate their absence (or lower abundance) in control, unpolluted sediments. In the environment, removal of sediments polluted with mercury, PCBs and PAHs in a location in the Baltic Sea resulted in the development of significantly different bacterial communities. Among the active members found after dredging, there was a decrease in the number of Deltaproteobacteria and Spirochaeta, and an increase of Gammaproteobacteria, which became dominant, Bacteroidetes and Alphaproteobacteria (Edlund & Jansson, 2006).

Summarizing the studies presented on chemical and heavy-metal pollution, we can conclude that, as in the case of nutrient enrichment, this type of perturbation again alters the composition of microbial communities (Fig. 3). But because this type of pollution selects for those microorganisms that are capable of degrading or chemically transforming the pollutants, and sometimes these organisms belong to particular phylogenetic groups, the studies conducted so far can point to particular types of bacteria (i.e. oligotrophic marine hydrocarbon-degrading Gammaproteobacteria, Actinobacteria, Sphingomonaceae or certain SRB phylotypes) likely to be contributing to alleviate the problem of chemical pollution in the environment. However, in many cases, these is only indirect evidence that does not fill the gap in knowledge on fundamental aspects of the functionality of microbial communities in chemically polluted environments. For example, novel or unrecognized microbial genera might be involved in the degradation of contaminants, most likely through metabolic cooperation with other microorganisms using degradation intermediates. Microorganisms as such, i.e. forming part of degradation networks, will not be isolated as typical pollutant degraders and therefore escape our knowledge, although they are probably key players in the environment. Other fundamental aspects, such as how pollution affects bacterial production rates, respiration rates or other microbial processes have not been analysed in chemically polluted environments, in contrast to what has been done in nutrient-enriched environments.

The risk of antibiotic discharges to the marine environment

The broad use of high amounts of antibiotics in intensive farming to prevent or treat animal infections is also causing environmental concern (see review Cabello, 2006 and references therein). The bactericidal action of antibiotics can cause changes in the composition of natural microbial communities by selectively inhibiting susceptible bacteria. Besides, exposure of natural communities to antibiotics might lead to the selection of resistant bacteria, which can then transfer their resistance determinants to opportunistic pathogenic bacteria in the environment. In relation to the marine environment, there are numerous data coming from aquaculture, where administration of high amounts of a variety of antibiotics in fish farms for prophylactic or therapeutic reasons is routinely done (Cabello, 2006; Sapkota et al., 2008). Antibiotic treatment in fish farms leads to increased concentrations of these compounds in water and sediments below the cages as well as in the fish stock produced. As a result, there is a positive selection for antibiotic-resistant bacteria in water and sediments from fish cages, surrounding areas and among fish-associated bacteria (Chelossi et al., 2003; Cabello, 2006; Sapkota et al., 2008). Often, the isolation of multiresistant bacteria (i.e. bacteria resistant to several antibiotics) is reported. This is in agreement with the simultaneous use of several types of antibiotics in fish farms (Dang et al., 2007; Sapkota et al., 2008). Antibiotic-resistant bacteria isolated from fish-farm water and sediments are diverse. Although most of them belong to the Gammaproteobacteria (genera Vibrio, Photobacterium, Pseudomonas, Pseudoalteromonas, Alteromonas, Citrobacter, Salmonella), isolates from the Alphaproteobacteria, Firmicutes, Actinobacteria and Bacteroidetes have also been reported (Furushita et al., 2003; Maki et al., 2006; Dang et al., 2007). The presence of antibiotic-resistance genes in these isolates, as well as the ability of some of the isolates for transferring the resistance genes by conjugation to Escherichia coli recipient strains, have been demonstrated (Rhodes et al., 2000; Furushita et al., 2003). Besides, closely related plasmids of the IncU group, carrying oxytetracycline-resistance determinants, have been found in isolates from hospitals in the UK and Germany and from fish-farm environments, demonstrating that plasmid transfer might occur between natural bacteria and potential human pathogens (Rhodes et al., 2000).

Instead of isolating antibiotic-resistant bacteria, Hargrave et al. (2008) followed a different approach to analyse antibiotic resistance. They measured profiles of resistance to the antibiotic oxytetracycline, widely used in aquaculture, in mixed bacterial communities from sediments around salmon aquaculture farms and in feed pellets (which were proposed as the source of antibiotics). Resistant bacteria able to grow between 40 and 160 μg mL−1 were detected in sediments within 100 m of salmon pens and in feed pellets. In contrast, in reference areas, there was growth inhibition at concentrations below 20 μg mL−1 of the antibiotic. Usually, numbers were higher in surface sediments but resistance levels were still high at intermediate depths, which mean that sediments around fish farms can constitute reservoirs for antibiotic-resistant bacteria in the marine environment (Hargrave et al., 2008).

It is important to recognize that fish farms are not the only source of antibiotics in the marine environment. Thus, oxytetracycline-resistant bacteria are present in surface seawater and sediments receiving the input of sewage treatment plants, as demonstrated in Jiaozhou Bay in China and Halifax Harbour in Canada (Dang et al., 2008; Hargrave et al., 2008). The study conducted in the area of VH, where discharges of 14.4 kg per day of 11 antibiotics from seven sewage treatment plants have been estimated (Minh et al., 2009), shows the magnitude of the problem. Anthropogenic use of coastal areas (i.e. beaches) also seems to have an effect on the natural abundance of antibiotic-resistant bacteria. Studies performed in sand from beaches of the Southern Baltic Sea coast showed that bacteria isolated from a recreational beach had higher resistance to antibiotics from different chemical families than those in nonrecreational areas (Mudryk, 2005; Mudryk et al., 2010). The level of multiresistance (resistance to several antibiotics) was also higher among isolates from the recreational beach.

The examples presented here provide evidence of the potential health risk derived from antibiotic contamination in the marine environment because they demonstrate the selection and enrichment of multiresistant bacteria. Some of them might be potential human pathogens (i.e. Vibrio, Pseudomonas, Salmonella) or might potentially transfer genetic determinants of resistance to pathogenic bacteria. But none of these studies have addressed how antibiotics alter the composition and functionality of microbial communities in marine environments. Given the magnitude and worldwide extension of the problem of antibiotic pollution, it is imperative to design carefully environmental and laboratory studies to address this issue, to determine which microbial groups are affected by the presence of antibiotics in the water and how antibiotics alter the functionality of microbial communities, both in water and in sediments.

Introduction of microbial pathogens into the marine environment

Several of the uses of coastal resources have the risk of introducing potential human pathogenic bacteria, as shown in Table 1. The highest risk is derived from direct sewage discharges, which are a source of human faecal bacteria. Therefore, sewage treatment and routine controls for faecal indicators (coliforms, E. coli, enterococci) in seawater in areas dedicated to bathing are usual procedures in developed countries (Stewart et al., 2008). However, sewage discharge is not the only source of faecal bacteria in marine environments. Other human-derived sources of potential pathogens are runoff from land (urban and agricultural areas), leaking septic tanks, sewer overflows, discharges from boats, etc. (Stewart et al., 2008). Sources of faecal bacteria other than human (i.e. birds, pets, wildlife, etc.) should not be disregarded (Choi et al., 2003; Dickerson et al., 2007), and therefore, determination of the human origin of the faecal indicators is important. A recent publication has presented a novel and interesting approach to assess faecal pollution in coastal watersheds using community-based indicators (Wu et al., 2010). Using PhyloChip, a phylogenetic microarray, these authors identified 503 operational taxonomic units (OTUs) characteristic of faecal samples (mainly belonging to the phyla Firmicutes, Proteobacteria, Bacteroidetes and Actinobacteria) which were designated as faecal subsample associated OTUs (FSAO). By analysing the similarity of bacterial communities in several environmental samples from two watersheds in Santa Barbara (CA) to FSAO, they could identify the samples exposed to faecal sources. The results obtained agree with traditional culture methods for the detection of faecal bacteria. In addition, they proposed a new community-based indicator to assess ecosystem health, based in the ratio of relative richness of three bacterial classes: Bacilli, Bacteroidetes and Clostridia to that of the Alphaproteobacteria (BBC : A). This ratio is higher in faecal and sewage samples and lower in samples not impacted by faecal material (Wu et al., 2010). Approaches like this represent an important step forward in the analysis of faecal contamination in coastal areas because it is based in powerful molecular methods, and in the assessment of whole communities instead of using a few (two or three) indicator bacteria.

Faecal material is also a source of human pathogenic enteric viruses. Viruses from faecal origin belong to the families Adenoviridae (adenovirus), Caliciviridae (i.e. Norwalk virus), Picornaviridae (i.e. poliovirus, hepatitis A) and Reoviridae (reoviruses and rotaviruses). These viruses cause a variety of diseases in humans, either by direct exposure to water or after ingestion of contaminated seafood (Griffin et al., 2003). Viruses are shown to persist better in the environment than the bacterial indicators used for water quality monitoring (Griffin et al., 2003); they persist better at lower water temperatures, associated to particles and in marine sediments and are rapidly inactivated by UV radiation (Griffin et al., 2003; Fong & Lipp, 2005). Apart from causing diseases in human and other marine mammals, epidemic viral infections can cause significant economic losses in aquaculture plants for commercial production of fish and shellfish (Lang et al., 2009).

In addition to discharges of faecal material, other human activities such as bathing constitute a source of potential microbial contamination. For example, enterococci and Staphylococcus aureus are transferred directly from the skin of bathers in high loads into seawater (3–6 × 105 and 6.1 × 106 CFU per person in 15 min exposure) or indirectly from sand carried on the body surface (Elmir et al., 2007, 2009). These results highlight the importance of nonenteric bacteria as potential pathogens in the environment (Stewart et al., 2008).

Besides the health risk posed by allochthonous microorganisms from faecal or other origin, there are several potential human pathogenic species that are indigenous to marine and estuarine environments, such as Vibrio cholerae, Vibrio vulnificus and Vibrio parahaemolyticus (Thompson et al., 2004; Stewart et al., 2008), which might be affected by human-derived activities. The first two species, V. chlolerae and V. vulnificus, can cause severe, fatal diseases. In contrast, V. parahaemolyticus infections are usually not life-threatening but they are very common worldwide, mainly due to the consumption of contaminated seafood (Collins, 2003; Baker-Austin et al., 2010). These opportunistic pathogenic bacteria have a variety of virulence factors such as haemolysins (V. vulnificus and V. parahaemolyticus), toxins (cholera toxin), colonization factors, etc. (Thompson et al., 2004). The presence of pathogenicity-associated genes from V. cholerae and V. parahameolyticus has been proven in environmental isolates from different Vibrio spp., suggesting that environmental vibrios might represent a reservoir of virulence genes in the environment (Lipp et al., 2002; Baffone et al., 2006). In the marine environment, vibrios are usually associated to surfaces and establish symbiotic relationships with phyto- and zooplankton (i.e. copepods), which might constitute vectors for disease transmission and for the development of disease outbreaks (Lipp et al., 2002; Thompson et al., 2004; Baffone et al., 2006). Therefore, conditions favouring the development of phytoplankton blooms, such as in eutrophicated environments, might favour the development of vibrios. In fact, as mentioned previously in this review, Vibrionaceae have been found in nutrient-rich environments (Yakimov et al., 2005; Zhang et al., 2007; Simonato et al., 2010) and antibiotic-resistant vibrios have been isolated from aquaculture plants (Furushita et al., 2003; Maki et al., 2006; Dang et al., 2007). In addition, vibrios proliferate at higher water temperatures and have a wide tolerance range to salinity, being able to survive well in low-salinity (brackish) water. Therefore, their incidence is expected to increase in a scenario of global climate change.

Human activities, mainly in coastal areas, also pose a potential risk for the development of harmful and toxic algal blooms or their occurrence at higher frequencies (Masó & Garcés, 2006; Kite-Powell et al., 2008). Different marine dinoflagellates (Dinophysis spp., Prorocentrum spp., Gymnodinium spp., Alexandrium spp., Pyrodinium spp., K. brevis and Gambierdiscus toxicus) and a diatom (Pseudonitzschia) produce toxins with a variety of neurologic, gastrointestinal, respiratory and irritating effects. Disease is usually caused by consumption of contaminated fish or shellfish, although aerosol inhalation and direct eye or skin exposure can also cause problems. In most cases, the symptoms are transitory, but ingestion of shellfish contaminated with saxitoxins or ciguatoxins might be fatal (Masó & Garcés, 2006; Kite-Powell et al., 2008). Environmental perturbations derived from human activities that might have an effect in the increased incidence of harmful algal blooms are multiple and include: nutrient enrichment (eutrophication), particularly in coastal waters; destruction of coastline due to coastal exploitation (i.e. increase of confined water bodies such as ports); risk of transporting active or resting cells and therefore potentially increasing the geographical range of the harmful species; and decreasing biomass of possible predators due to overfishing or changes in their environment (Masó & Garcés, 2006).

Potential pathogens might be disseminated in the marine environment due to long-distance transport and discharge of ship ballast water (used for ship stability and trim). For example, ballast water has been shown to contain epidemic-causing serotypes of V. cholerae, such as O1 and O139 (McCarthy & Khambaty, 1994; Ruiz et al., 2000; Aguirre-Macedo et al., 2008), as well as other potential human pathogens such as faecal coliforms, E. coli (including strain O157), Enterococcus spp. (McCarthy & Khambaty, 1994; Aguirre-Macedo et al., 2008) or harmful algae (Masó & Garcés, 2006). Concerns of the danger of ballast water discharges, not only for the spread of microbial pathogens but also of invasive species, have resulted in the establishment of guidelines for ballast water management and the promulgation of the International Convention for the Control and Management of Ship's Ballast Water and Sediments in 2004 by the International Maritime Organization.

The amount of total bacteria and virus-like particles in ballast water is in the order of 108–109 and 109–1010 L−1, respectively (Ruiz et al., 2000; Drake et al., 2001). These numbers are highly variable and depend on many factors such as the source region of the boat, season of the year and ballast-water management, i.e. whether or not there had been open-ocean exchange of water (Drake et al., 2001). The changes in the microbial component of ballast water during a transoceanic voyage in tanks with and without open-ocean exchange have been studied by quantifying bacterial and viral numbers (Drake et al., 2002). The results showed that containment of ballast water in boat tanks led to a decrease in the bacterial and viral load in comparison with the initial water, irrespective of performing open-ocean exchange or not (Drake et al., 2002). Bacterial communities in ballast water initially resemble those of the source seawater, which is usually from a coastal location (Tomaru et al., 2010). This ballast water is replaced in the open ocean at some point during the voyage to reduce the risk of discharging invasive species in the reception port. The open-ocean water filling the tanks has a significantly different microbial composition (Tomaru et al., 2010). This means that allochthonous microorganisms from coastal regions are routinely discharged into the open ocean, and vice versa, open-ocean microorganisms are discharged into coastal regions. However, the microbial assemblages discharged in each case are different from those in the source water because there is a change in composition during the voyage within the time frame of days (Tomaru et al., 2010). A recent study analysed bacterial diversity in ballast water in a ship anchored in Xiamen Port, and compared it with that of the receiving harbour seawater (Ma et al., 2009). The source of the ballast water was Singapore and it had been partially replaced with water from the South China Sea. The analysis of the corresponding 16S rRNA gene clone libraries revealed that ballast water contained a less diverse bacterial community, with representatives of only two classes, Alpha- and Gammaproteobacteria. There was no evidence of the presence of pathogenic bacteria. Within these two groups, the phylotypes retrieved in ballast water were different from those in Xiamen Harbour seawater (Ma et al., 2009). Thus, wherever it is done, the discharge of ballast water might potentially alter, at least transiently, the autochthonous microbial composition of the discharge site. Whether this is significant or not has not been determined. If an open-ocean exchange is performed, we can expect that most of the coastal microorganisms will not be able to compete with the autochthonous bacteria in the more oligotrophic open ocean, and the opposite, i.e. oligotrophic bacteria from the open ocean will not survive in coastal regions. Results from metagenomic studies show that microbial communities with dissimilar genomic composition (i.e. genetic repertoire) develop in different marine locations or at different depths (DeLong et al., 2006; Rusch et al., 2007). Therefore, the success of an invasive microorganism, potentially dangerous such as a pathogen, would depend on the size of the inoculum, its capability to survive in the new environment and its competitive ability in facing a microbial community adapted to the conditions at the site of discharge.

Emerging threats in the context of global warming and climate change

As reported by Halpern et al. (2008), the alteration of parameters related to climate change are among the most important stressors to marine ecosystems. Thus, the interest in studying the effect of processes such as increase of seawater temperature and ocean acidification is increasing rapidly in the last years. No particular studies have been done targeting the microbial compartment but the information gathered so far provides clues on how microbial communities could be affected.

Ocean acidification and temperature increase

Anthropogenic activities such as the burning of fossil fuels, deforestation, industrialization and cement production are causing increased levels of atmospheric carbon dioxide (CO2) and, consequently, an increase in dissolved CO2 in the oceans. This process alters the chemistry of seawater and is usually referred to as ocean acidification due to its effects in reducing water pH (for a review, see Guinotte & Fabry, 2008). The acidification of the oceans will be detrimental for calcifying organisms and will cause changes in species distribution and abundance, as well as in food-web dynamics and structure. The increase in dissolved CO2 has an effect in the carbon cycle of the oceans but also in the cycles of the major nutrient elements: nitrogen, phosphorus, silicon and iron (for a review, see Hutchings et al., 2009), and therefore, it will affect important microbial processes in the marine ecosystem. The information available indicates that three important processes in the nitrogen cycle might be affected. For example, nitrogen fixation is predicted to increase in a high-CO2 ocean, based on the results of experiments done with nitrogen-fixing Trichodesmium and unicellular Cyanobacteria. In contrast, a decrease in pH seems to reduce nitrification rates and the abundance of bacterial and archaeal nitrifiers in seawater (Hutchings et al., 2009). Consequently, the fluxes of oxidized nitrogen species, such as nitrate, will be affected. This in turn would have a negative effect on denitrification rates, although there can also be positive effects over this process due to the expansion of suboxic zones in the oceans. An increase in the relative proportions of ammonium to nitrate due to decreased nitrification will cause a shift in the composition of primary producers, by favouring microbial components such as picocyanobacteria and nanoflagellates (Hutchings et al., 2009), affecting the structure of the microbial food web.

On the other hand, global warming will lead to an increase in seawater temperatures and cause a stronger stratification of the oceans. This will limit the fluxes of nutrients from sediments and the deep sea (i.e. phosphorus, iron), and potentially intensify problems of hypoxia and anoxia. Changes in the climate system can also cause alterations in the abundance and global distribution (i.e. spread from tropical areas to lower latitudes) of pathogens such as Vibrio spp., with the consequent risk of increasing the incidence of Vibrio-related diseases. The abundance of V. cholerae, the causative agent of cholera epidemics, is already shown to follow climatic patterns (Lipp et al., 2002; de Magny et al., 2008). Higher water temperatures, together with increased nutrient levels, would also increase the frequency and severity of harmful algal blooms (Moore et al., 2008). Environmental changes due to anthropogenic activities (seawater warming or increased CO2 and nutrient supply) are also suspected of being involved in the expansion of coral diseases worldwide, diseases in which microorganisms are important (Sokolov, 2009). In addition, strong rainfall events and flooding due to sea-level changes, both expected under a global change scenario, would result in an increase of estuarine and brackish (low-salinity) environments, which again would favour the growth of Vibrio spp. (Lipp et al., 2002; Baker-Austin et al., 2010). Moreover, floods may also cause problems of inefficient water sanitation and increase runoff from land, with the consequent increase in the risk of faecal (bacterial and viral), nutrient and chemical contamination of the marine environment (Kite-Powell et al., 2008).

Carbon sequestration by stimulation of the biological pump

The possibility of generating phytoplankton blooms by ocean fertilization with the aim of increasing the flux of organic matter towards the deep ocean (causing carbon burial and alleviating problems of increasing CO2 concentrations) is catching the attention of companies for commercial exploitation. This represents a new anthropogenic use of the marine environment. Several scientific experiments involving iron and phosphorous fertilization have been performed in order to test the premise that these nutrients were limiting primary production in different oceanic regions (Thingstad et al., 2005; Boyd et al., 2007). However, based on the experience gathered on the functioning of marine ecosystems and the interplay of biogeochemical cycles, there is strong scientific criticism of the usefulness of ocean fertilization to sequester carbon, as well as concerns about the risks associated (Secretariat of the Convention on Biological Diversity, 2009). In particular, the plans of the Australian company Ocean Nourishing Corporation to discharge urea in nitrogen-deficient areas of the ocean have been criticized on scientific grounds (Glibert et al., 2008). Several important risks have been pointed out such as the alteration of nutrient stoichiometry of the water, eutrophication, stimulation of photosynthetic species with high-buoyancy (low sedimentary) properties such as cyanobacteria and dinoflagellates which are preferential users of urea as nitrogen source, production of blooms of toxic dinoflagellates and the development of hypoxic/anoxic zones, which in turn might increase the release of other greenhouse gases such as methane and nitrous oxide (Glibert et al., 2008). In this scenario, there will not be an increase in CO2 sequestration and carbon burial, and in addition, undesired gases will be released into the atmosphere, aggravating the problem instead of alleviating it.

Concluding remarks

Worldwide marine ecosystems suffer from the impact caused by human activities. In view of the development of our societies and the increase in human population, the stress posed by humans to the marine environment will continue to increase. The microbial component of marine ecosystems has been neglected in many studies of anthropogenic impact. Therefore, our knowledge on microbial communities of anthropogenically impacted environments lags far behind those involving higher organisms. Many studies report changes in prokaryotic numbers and a few have included measurements of microbial activity. Changes in diversity have also been analysed, although in many cases, this has been done only by electrophoretic profiling methods. As a result, we have a partial view on the composition of microbial communities in stressed environments. Therefore, there is a need for a better cataloguing of microbial diversity in anthropogenically stressed environments. New technologies such as remote ocean-observing systems, metagenomics, metatranscriptomics and massive tag sequencing can help us to gain more information on these communities. Thanks to the use of comparative studies of impacted and reference sites, we know that human impact causes significant changes in microbial community composition. However, we are still unable to make predictions on how important this could be for the functioning of the ecosystems, particularly in the long term, especially if the stressor does not disappear or it is likely to increase. For example, we know already that certain microbial groups are favoured in conditions of nutrient enrichment or chemical pollution, but most of this information is provided for broad phylogenetic groups (i.e. division, class) and not for particular species or ecotypes. Therefore, there is an urgent need for delineating patterns of microbial occurrence/absence/abundance in human-impacted environments at higher levels of phylogenetic resolution (i.e. phylum, class, genera, species, etc.). There is also a need for relating the changes in diversity to the metabolic processes, which are important for the functioning of the ecosystem. Because of the complexity of marine ecosystems, these goals would need the combination of environmental observations as well as carefully designed microcosms or mesocosms experiments. In the context of global climate change, for which human activities can be blamed, microbial communities of human-stressed environments are of paramount importance, and therefore the study of their composition, dynamics and functioning is of the utmost relevance for the development of less destructive human practices or for alleviating problems already present in our environment.

Acknowledgements

The authors wish to thank Margarita Gomila for her comments on the manuscript. Research of the group is supported by grant CTM2008-02574/MAR from the Spanish Ministry for Science and Innovation (MICINN) with FEDER cofunding. M.P.L. and J.M.P.-V. are supported by PhD fellowships of MICINN.

Ancillary