Introduction

Coral reefs have flourished in tropical shallow water regions for over 240 million years (Stanley and Fautin 2001). However, during the last 30 years coral reefs have faced severe threats due to anthropogenic climate change, crown of thorns starfish (COTS), disease, overfishing and pollution (Hoegh-Guldberg 2011; De‘ath et al. 2012). For instance, the emission of carbon dioxide (CO2) leads to a decrease in the ocean’s pH and adversely affects calcium carbonate deposition by calcifying organisms such as environmentally susceptible coral species (Hoegh-Guldberg et al. 2007; Putnam et al. 2016). Increased ocean surface temperature as a consequence of global warming, not only result in more frequent and severe tropical storm events (reviewed by Knutson et al. 2010) but also pushes the coral-Symbiodinium symbiosis towards its thermal tolerance limit (Hoegh-Guldberg 1999). Once the temperature exceeds the resistance threshold of the symbiosis, the interaction between corals and their photoautotrophic Symbiodinium breaks down and corals bleach (Brown 1997). Mass coral bleaching events have occurred more frequently over the last decades (Donner et al. 2005; Hoegh-Guldberg et al. 2007; Baker et al. 2008). The most recent global bleaching episode in 2015–2016 was the longest bleaching event recorded in history (US National Oceanic and Atmospheric Administration) and has led to severe bleaching worldwide. In the northern sector of the Great Barrier Reef (GBR) in Australia, this bleaching event caused the mortality of up 35% of all corals (GBRMPA 2016; Normile 2016; Hughes et al. 2017). In addition to the hazards emerging from global change, localized reef-scale anthropogenic impact can further influence the ecological resilience of corals and other reef organisms to global stressors (Knowlton and Jackson 2008). For example, overfishing has led to an alteration of the trophic food web (Valentine and Heck 2005). With the removal of herbivorous fish from coral reefs, macroalgae and turf algae are less efficiently grazed. This can result in a phase shift from coral to macroalgae dominated reefs (Burkepile and Hay 2006; Vermeij et al. 2010). Changes in land-based practices can also impact reef ecosystems with land-clearing, coastal urbanization and intense agricultural practices contributing to decreased water quality on reefs, which in turn further contributes to declining reef health (De‘ath and Fabricius 2010).

Monitoring programs, such as ‘Reef Check’ and ‘The Global Coral Monitoring Network’ use visual surveys of reef associated fish, invertebrates, substrate composition, reef damage and disease impacts to assess reef health (Hodgson 2001; Hill and Wilkinson 2004). Most visual cues (e.g. tissue necrosis, mortality and shifts in community composition) only become evident in the advanced stages of coral reef ecosystem stress, when ecosystem health and resistance are already compromised. At this stage the ability to recover is dependent on the resilience of the ecosystem and the success of management interventions is jeopardized. Hence, the development of a reliable early warning system that facilitates management intervention before severe damage occurs is clearly warranted.

Microorganisms are key drivers of large-scale biogeochemical cycles in the oceans (Azam et al. 1983; Whitman et al. 1998; Falkowski et al. 2008) and also play a fundamental role in coral reef ecosystem functioning, through mediating nutrient cycling (Charpy et al. 2012; Tout et al. 2014). Furthermore, microbes live in intimate relationships with benthos-dominating life forms such as corals, sponges and macroalgae, where they have a vital role in host fitness through additional nutrient provision, removal of waste products and the exclusion of opportunistic microbial pathogens (Egan et al. 2013; Blackall et al. 2015; Bourne et al. 2016; Webster and Thomas 2016). However, disturbance events can alter the natural microbial community structure, abundance and metabolic functions either directly or indirectly (Shade et al. 2012). Disturbance-related deviations from the naturally occurring microbial communities may provide useful indications for coral reef ecosystem stability and facilitate sensitive predictions of environmental stress (Fig. 1).

This review aims to assess the utility of microorganisms as a diagnostic tool for assessing water quality and climate-driven stress in the coral reef ecosystem. Insights are provided into (1) the implementation of biological indicator approaches; (2) the potential role of microbial indicators in coral reef monitoring programs; (3) the function and diversity of microbes associated with dominating benthic life forms (corals, sponges and macroalgae) and their surrounding environments (seawater and sediment) and (4) how state-of-the-art methods used to study microbial community composition and function can be implemented into diagnostic tools to assess coral reef ecosystem health.

Fig. 1
figure 1

Microbial indicator approach to assess coral reef health. Coral reefs are exposed to increased environmental pressures emerging from local (e.g. pollution, overfishing) and global impacts (e.g. rising sea surface temperature). This facilitates a shift from coral to algae dominated reef systems [adapted from Sandin et al. (2008)]. Accompanied by the increase in environmental pressures, the microbial community composition and function associated with coral reefs change along the gradient of disturbance from a beneficial and commensal microbiome towards microbial communities dominated by opportunists and pathogens. Overall compositional and functional changes of microbial communities associated with seawater, sediment and habitat forming taxa (corals, sponges and macroalgae) but also the occurrence or loss of specific microbial taxa/function can provide useful indications for the prevailing environmental condition. The application of microbial indicator taxa, function and/or community assemblages will allow for a rapid prediction of environmental disturbance and the health state of a coral reef. This enhanced predictive capability is paramount to efficiently monitor coral reef health and locally manage environmental pressures such as water quality

Biological indicators: definition, function and application

Biological indicators, also referred to as indicator species, are organisms that can be used to monitor habitat conditions and environmental changes based on their niche preferences (McGeoch and Chown 1998; De Cáceres et al. 2010). The presence/absence or abundance values of selected organisms or communities at specific sampling sites are monitored, allowing conclusions to be drawn on the ecological integrity of a certain ecosystem (Carignan and Villard 2002). Depending on the response time, bioindicators can either be used to detect severe short-term stress events or to better understand the long-term effect of chronic stress exposure (Cairns and Pratt 1993). The use of taxa with moderate tolerance towards environmental variability is favored against rare or ubiquitous taxa, which respectively show high sensitivity or high tolerance towards environmental perturbations (Holt and Miller 2011). Cooper et al. (2009) identified the following five selection criteria, to be considered when choosing bioindicators: specificity, monotonicity, natural variability, practicality and ecological relevance (see Table 1).

Table 1 Summary of criteria for the selection of microbial indicators to monitor water quality changes and predict stressors in coral reef ecosystems

Kolkwitz and Marsson (1908) developed one of the first indicator approaches to evaluate the degree of pollution in rivers based on the occurrence of saprobic micro- (e.g. ciliates and flagellates) and macroorganisms (e.g. insect larvae, molluscs, bivalves, annelids and crustaceans). Since then, various biomonitoring programs have been developed and successfully applied in estuarine and freshwater ecosystems by assessing indicator species within the macrobenthic invertebrate community e.g. AUSRIVAS (Australian River Assessment Scheme; Smith et al. 1999), RIVPACS (River Invertebrate Prediction and Classification System; Wright 1995), SIGNAL (Stream Invertebrate Grade Number Average Level; Chessman 1995) and WFD (European Union Water Framework Directive; European Parliament 2000). Biomonitoring approaches also find application in coral reef health monitoring (reviewed by Cooper et al. 2009). For example, the FORAM index (Hallock et al. 2003) provides insights into the water quality of coral reefs by quantifying size classes of Foraminifera in the upper sediment layer.

Although indicator approaches are frequently used in conservation biology, land management, landscape mapping and in the design of protected areas, standardized approaches and statistical methods to identify suitable indicators are still scarce. Dufrêne and Legendre (1997) developed the Indicator Value Analysis (IndVal) as a statistically valid method for determining indicator species and species assemblages. IndVal considers the specificity, which is the mean number of individuals of a species across sampling sites of a certain habitat in comparison with other habitats, and the fidelity, described as the relative frequency of occurrence of a species in the sampling sites of a specific habitat. The IndVal analysis represents an efficient method to identify indicator organisms and can provide critical information on the ecological integrity of an ecosystem (McGeoch and Chown 1998; Gardner 2010). The IndVal approach has found application in various terrestrial and aquatic studies to assess habitat quality based on both macro- (McGeoch et al. 2002; Muotka et al. 2002; Bazelet and Samways 2011) and microorganisms (Auguet et al. 2010; Fortunato et al. 2013; Glasl et al. 2016; Ziegler et al. 2017).

The integration of biological indicators into ecosystem monitoring programs can provide significant advantages. For example, by focusing on a targeted group of species, monitoring can become more cost and time efficient. Furthermore, biological indicators can provide early warning of environmental stress and might directly reveal the cause rather than simply the existence of a disturbance event. However, the choice of appropriate species or species assemblages is crucial for the effectiveness of bioindicators and hence, the selection should be based on sound quantitative approaches including consideration of the natural in situ variability (Carignan and Villard 2002).

Microorganisms as bioindicators to assess ecosystem health

The potential value of microorganisms for efficiently monitoring ecosystem health remains largely unexplored, despite the fact that microbes show fast, specific responses to environmental perturbations (Teeling et al. 2012; Wemheuer et al. 2015; Haas et al. 2016) which are the desirable characteristics of bioindicators (Table 1). One area where microorganisms have been exploited as indicators is in monitoring contamination of water supplies by coliform counts associated with feces from warm-blooded animals (Ashbolt et al. 2001). Coliform counts are now used to monitor and manage drinking water supplies and recreational water qualities worldwide (Boehm and Sassoubre 2014), successfully enabling the prevention of numerous human gastrointestinal illnesses (Wade et al. 2010). In coral reef environments, coral mucus has been shown to be more efficient in trapping fecal indicator bacteria and human enteric viruses than the surrounding seawater; hence, coral mucus provides an improved medium to monitor sewage contamination (Lipp and Griffin 2004). Furthermore, seagrasses have recently been shown to reduce the abundance of microbial pathogens entering nearby coral reef and mitigate the disease risk in associated coral populations (Lamb et al. 2017).

Intensive research on the structural and functional capacity of the human microbiome has revealed its enormous contribution to pathogenesis and immune system modulation of the host, as well as its influence on host development and physiology (Arrieta and Finlay 2012; Eloe-Fadrosh and Rasko 2013; Kostic et al. 2015; Ghaisas et al. 2016). This knowledge has revolutionized our current understanding of host–microbe interactions and has led to the development of diagnostic and therapeutic approaches targeting the human microbiome (Reardon 2014). For example, transplantation of faecal microbiomes has successfully cured humans affected by recurrent Clostridium difficile infections and microbiome transplantation has been suggested as a therapeutic cure for inflammatory bowel disease and obesity (Gupta et al. 2016). In the same way, it is feasible that microbiome manipulation and probiotic treatment could be used to increase the health and tolerance of other reef based host-associated systems (Reshef et al. 2006; van Oppen et al. 2014; Webster 2017). Also, similar to the approach in humans, the identification of imbalances in the microbial communities (dysbiosis) at the ecosystem level can facilitate diagnostic interpretations of environmental health. For example, a bacterial community based index has recently been developed to assess the ecological status of estuarine and costal environments (Aylagas et al. 2016). Another recent example applied to coral reef ecosystems has been the development of microbialization scores, which attempt to assess human impacts on coral reefs based on the metabolic rates of microbial communities and reef-associated fishes (McDole et al. 2012). While microbialization has been shown to occur on a global scale (Haas et al. 2016), additional research is needed to adopt the approach to individual reef ecosystems. Microbial monitoring has also recently been introduced to the monthly sampling program of the National Mooring Network of IMOS (Integrated Marine Observing System), which targets oceanographic phenomena in Australian coastal waters (IMOS 2016). Increased appreciation of microorganisms in host and/or ecosystem health, together with recent advances in molecular techniques, now allow for detailed in situ investigations of the microbial community structure and functions and for standardized and efficient data analysis.

Changes in microbial communities due to disturbance may directly affect ecosystem processes. Therefore, it has been suggested to include microbial community composition (e.g. 16S ribosomal RNA gene sequencing) into process models that predict ecosystem responses to global change (Allison and Martiny 2008). Microbial functional redundancy is also of particular interest since metabolic capabilities can be decoupled from the phylogenetic position of microorganisms due to convergent evolution, gene loss or horizontal gene transfer (Ochman et al. 2000; Martiny et al. 2013). Phylogenetic diversity is thought to positively affect the stability of an ecosystem, as it increases the probability that complementary functional traits are present (Yachi and Loreau 1999). However, ecosystem processes can remain constant after disturbances even when a compositional shift has been observed (Wohl et al. 2004; Allison and Martiny 2008; Banerjee et al. 2016). Hence, functional approaches (e.g. metagenomics, metatranscriptomics) are also required for monitoring and predicting ecosystem changes. Functional and compositional changes can be assessed based on the entire community or on selected microbial indicator taxa.

Microbial life in the coral reef ecosystem

Within coral reefs, microorganisms colonize various habitats including the water column, the sediment and the benthic community, such as corals, sponges and macroalgae (Moriarty et al. 1985; Friedrich et al. 1999; Rohwer et al. 2001; Hewson and Fuhrman 2006; Barott et al. 2011; Bourne and Webster 2013). The enormous complexity of coral reefs and their associated microbial communities has resulted in studies focused on these specific compartments; however, these habitats should not be considered as isolated from each other but rather seen as parts of a single ecosystem with a strong benthic-pelagic exchange (Lesser 2006; Garren and Azam 2012). Holistic approaches that combine the different reef habitats are urgently required to better understand the function and contribution of microorganisms to reef health and resilience.

Microbial diversity and function in coral reef waters

Microorganisms are moderately abundant (average densities of between 3 and 9 × 105 cells ml−1) in nutrient-poor coral reef waters, where they play diverse roles related to nutrient cycling that ultimately affect the entire reef ecosystem (Sorokin 1973, 1978; Ducklow 1990; Gast et al. 1998). For instance, photoautotrophic picophytoplankton significantly contributes to the biomass and primary productivity of oligotrophic reef waters (Stockner 1988). Additionally, as part of the marine microbial-loop, heterotrophic bacteria utilize dissolved organic matter (DOM) in the water column, establishing an important recycling step that makes energy available to higher trophic levels (Azam et al. 1983). In marine environments DOM is primarily of phytoplankton origin, as 2–50% of the photoautotrophically fixed carbon leaks into the water column (reviewed by Thornton 2014). Coral mucus, fueled by the photosynthetic activity of the coral’s algal symbionts, additionally contributes to the DOM pool of coral reef waters (Wild et al. 2004a). The fixation of inorganic nitrogen is another key microbial function in coral reef environments, where nutrients are scarce and thus limit growth (Charpy et al. 2012).

Processes shaping the microbial communities associated with coral reef waters vary over space and time. Distinct microbial communities along different niches within coral reef waters are the result of habitat structure, the presence of benthic host organisms and local biogeochemical conditions (Tout et al. 2014). Seasonal effects, such as temperature, rainfall and water quality affect microbial community composition in shallow water reef sites (Angly et al. 2016). Anthropogenic impact (e.g. land-use and fishing) also affects coral reef microorganisms with disturbed reefs possessing higher microbial abundances and a larger proportion of microbial taxa related to potential pathogens (Dinsdale et al. 2008). The replacement of macro- with microorganisms under increased human influence has been referred to as microbialization (Jackson et al. 2001). Based upon the microbialization concept, McDole et al. (2012) developed the ‘microbialization score’, a metric to assess the level of human impact and compare the health of coral reefs across time and space.

Microbial diversity and function in coral reef sediments

Coral reef sediments are typically dominated by calcareous sand, characterized by high permeability, porosity and surface area (Rasheed et al. 2003). This enables large numbers of microorganisms to settle and grow on the sediment grains with prokaryotic abundance estimated at between 1 and 2 × 109 cells cm−2 of sediment surface (Wild et al. 2006). In addition, the microbial communities associated with reef sediments are highly diverse, with vertical community stratification caused by redox gradients (Rusch et al. 2009). The oxygenated upper sediment layer is dominated by heterotrophic processes and provides an important recycling step for dissolved and particulate organic matter (Wild et al. 2004a). Wild et al. (2004b) demonstrated that sloughed coral mucus acts as an efficient particle trap in the water column and rapidly carries nutrients to reef sediments, where diverse microbial metabolisms successfully remineralize them. Fixation of inorganic nitrogen is also mediated by bacteria and archaea present within coral reef sediments and can substantially contribute to the overall coral reef nitrogen budget (Cardini et al. 2014).

Importantly, Uthicke and McGuire (2007) identified clear bacterial community differences in surface sediments collected from inshore and offshore locations on the GBR and proposed they be used as biological indicators for water quality. Furthermore, a study conducted in the Red Sea has shown a nutritional link between seasonal dynamics and sediment-associated bacterial communities (Schöttner et al. 2011). In addition to their metabolic functions, sediment associated microbes may also act as a seed-bank for microbes associated with coral mucus (Carlos et al. 2013; Glasl et al. 2016), another example of the tight microbial connections established within the coral reef ecosystem.

Corals and their microbiomes

Coral holobionts (Rohwer et al. 2002) comprise an array of microorganisms including fungi, endolithic algae, bacteria, archaea and viruses (Ritchie and Smith 1997; Sharshar et al. 1997; Rohwer et al. 2002; Koren and Rosenberg 2006; Thurber et al. 2008). The coral host itself provides several microhabitats for its microbial associates, such as the surface mucus layer (Rohwer et al. 2002; Frade et al. 2016), the tissue (Bourne and Munn 2005), the skeleton (Sharshar et al. 1997) and the gastrovascular cavity (Herndl and Velimirov 1985). Each of the microhabitats within a coral colony is associated with a distinct microbial community (Rohwer et al. 2002; Sweet et al. 2011; Bourne et al. 2016).

Bacterial communities associated with corals are clearly distinct from the bacterial communities in the surrounding environment (Frias-Lopez et al. 2002) and bacterial cell abundance within coral mucus is also 10-fold higher than in seawater (Garren and Azam 2010). Each coral species serves as a unique habitat and is associated with a specific microbial community (Sunagawa et al. 2010), with some species maintaining stable microbiomes over large geographic scales (Rohwer et al. 2002), suggesting that the coral host plays a key role in structuring its bacterial community.

To deal with the enormous diversity of microorganisms associated with corals, and separate beneficial and opportunistic bacteria and archaea from the bulk microbial community, understanding coral holobiont stability and functionality is paramount. For instance, the mucus microbiome of healthy Porites astreoides colonies is dominated by Endozoicomonadaceae, whose loss is coupled to deterioration in holobiont health (Meyer et al. 2014; Glasl et al. 2016). Endozoicomonas also dominate the microbiome of Stylophora pistillata and Pocillopora verrucosa tissues; Endozoicomonas genotypes vary over geographic space in S. pistillata, whereas in P. verrucosa the genotype of these endosymbionts remains the same. The genotype specificity may relate to differences in the life history strategies of corals (Neave et al. 2017). Based on meta-analysis of the core microbiomes of diverse coral species, members of Actinobacteria were recently identified as ubiquitous symbionts of corals (Ainsworth et al. 2015). A decrease of Actinobacteria in coral microbiomes during periods of temperature and algal stress further suggests their beneficial role in the coral holobiont (Zaneveld et al. 2016).

Potential functions of the coral’s microbiome vary from nutrient supply and recycling (Lesser et al. 2004, 2007) to protection against pathogens (Rohwer et al. 2002; Ritchie 2006; Shnit-Orland and Kushmaro 2009; Raina et al. 2016). For example, the depletion of the coral’s beneficial mucus microbiome affected holobiont health under in situ conditions (Glasl et al. 2016). This suggests that major disturbances of the surface mucus microbiome may open a niche for potentially opportunistic and/or pathogenic bacteria, which can further lead to diseases or host mortality. Recent investigation of the functional capabilities of the bacterial and archaeal community associated with corals revealed carbon fixation and degradation pathways and the presence of genes involved in sulfur and nitrogen cycling (Wegley et al. 2007; Siboni et al. 2008; Kimes et al. 2010; Yang et al. 2013; Rädecker et al. 2015). Nitrogen fixation capabilities are of particular interest as coral reefs are nitrogen-limited ecosystems (Falkowski et al. 1993). Until recently, Cyanobacteria were thought to be the primary suppliers of fixed nitrogen in the coral holobiont (Lesser et al. 2004, 2007). The discovery of the ubiquitous association with highly host specific diazotrophs (nitrogen fixing microorganisms) suggests their important functional role in the nitrogen supplementation of the coral holobiont (Siboni et al. 2008; Lema et al. 2012, 2014), especially when external nitrogen sources are limited (Cardini et al. 2015). Nitrifying, denitrifying and ammonia oxidizing microbes have also been described to be associated with corals; however, their precise functions within the holobiont remains understudied.

The microbiome of corals has been described as persistent over space and time (Rohwer et al. 2002). However, compositional and functional shifts have been observed in association with seasonal environmental variation (Li et al. 2014), host mucus-shedding dynamics (Glasl et al. 2016), thermal stress (for example coral bleaching; Bourne et al. 2008), disease (Séré et al. 2013; Meyer et al. 2014), coral’s proximity to macroalgae (Barott et al. 2012; Sweet et al. 2013) and increasing environmental pollution (Garren et al. 2009; Kelly et al. 2014; Ziegler et al. 2016). For example, thermal stress leads to destabilization of coral-algae interactions (Brown 1997) and a shift towards an opportunistic and/or pathogenic microbial community (Ainsworth et al. 2008; Bourne et al. 2008; Littman et al. 2011). Water quality changes associated with increased sediment and nutrient run-off can increase microbial abundance (Dinsdale et al. 2008; D’Angelo and Wiedenmann 2014), coral disease frequency (Bruno et al. 2003; Thurber et al. 2014) and higher macroalgal abundance on coral reefs (Kline et al. 2006). Macroalgae exudates stimulate bacterial activity and copiotrophic bacterial growth, and cause shifts in the coral microbiome attributed to increased labile DOC and toxic secondary metabolites (Morrow et al. 2011; Barott et al. 2012; Vega Thurber et al. 2012; Nelson et al. 2013; Sweet et al. 2013; Haas et al. 2016). Environmentally induced changes in the coral microbiome generally result in higher microbial abundance and a shift away from beneficial microbes towards opportunistic and/or pathogenic bacterial taxa, such as Vibrionaceae and Rhodobacteraceae (Vega Thurber et al. 2009; Bourne et al. 2016; Rothig et al. 2016; Ziegler et al. 2016). Increased nutrient run-off in combination with reduced grazers (hence higher algal abundance) destabilizes the coral microbiome with detrimental consequences for the host, particularly when exposed to additional stressors such as parrotfish bites and thermal stress (Zaneveld et al. 2016). The resulting imbalance in the holobiont composition (dysbiosis) can lead to functional changes of the microbiome and facilitate disease development or alterations in metabolism and/or immunity that lead to bleaching and/or necrosis, and ultimately coral death (reviewed by Thompson et al. 2015).

Sponges and their microbiomes

Marine sponges (phylum: Porifera) are a highly diverse component of coral reefs, usually exceeding the diversity of corals and algae (Diaz and Rützler 2001). Sponges have the ability to filter up to 50,000 times their own volume every day (Reiswig 1971a) and due to this active suspension feeding they play a key role in benthic-pelagic coupling, thus providing a vital trophic link between the benthos and the ambient seawater (Gili and Coma 1998; Southwell et al. 2008). Sponges are primarily considered to feed on picoplankton (0.2–2 µm) (Reiswig 1971b; Pile et al. 1997; Hanson et al. 2009); however, several recent studies have shown that certain sponges are also able to assimilate dissolved organic matter (DOM) from the water column (De Goeij et al. 2008; Mueller et al. 2014; Rix et al. 2016). Cryptic Caribbean sponges, for instance, transform DOM into particulate organic matter (POM) and thereby significantly contribute to the recirculation of nutrients in coral reefs, a phenomenon known as the sponge-loop (De Goeij et al. 2013), which is likely mediated by microbes living within the sponge.

Sponges generally live in close association with a wide variety of microorganisms including bacteria, archaea, unicellular algae and protists. These microorganisms are often present in high abundance, accounting for up to 40–60% of the total sponge volume and reaching cell abundances that exceed those in the surrounding seawater by several orders of magnitude (Hentschel et al. 2006). Sponges with high microbial densities are referred to as ‘high microbial abundance sponges’ (Vacelet and Donadey 1977; Hentschel et al. 2003), in contrast to marine sponges that harbor only a relatively small number of microorganisms and are referred to as ‘low-microbial abundance sponges’ (Hentschel et al. 2003). The implications of these diversity differences for host resistance and resilience remain elusive. In addition to the importance of microorganisms as a sponge food source (Reiswig 1971b; Pile et al. 1996), microbial associates have also been reported to participate in a diverse range of interactions, including parasitism, commensalism and mutualism (reviewed by Taylor et al. 2007) with significant implications for the ecology, biology and physiology of sponges (Thacker and Freeman 2012; Bourne and Webster 2013).

Microbes are generally located in the sponge mesohyl region, a layer of connective tissue where microbial cells are either freely occurring or enclosed within specialized bacteriocyte cells (reviewed by Taylor et al. 2007). The mesohyl is also the place where phagocytosis of food particles (e.g. picoplankton) takes place; hence, host-symbiont recognition mechanisms must be in action to prevent the phagocytosis of symbiont cells (Wilkinson et al. 1984; Wehrl et al. 2007; Nguyen et al. 2014). Furthermore, archaea and Cyanobacteria have been found in the pinacoderm, the outer surface of sponges formed by epithelial cells (Wilkinson 1980; Webster et al. 2001). Sponge-symbionts generally show high host species-specificity and only a few bacterial species seem to be ubiquitously present in sponges. Bacterial species associated with different sponges are more closely related to each other than to bacteria from the ambient water column and, hence, these monophyletic groupings of sponge symbionts are often referred to as ‘sponge-specific sequence clusters’ (SCs) (Hentschel et al. 2002; Schmitt et al. 2012). Whereas about half of the SCs have been shown to be vertically transmitted from adult sponges directly to their offspring, the other half is thought to be horizontally acquired by each generation from the surrounding seawater (Webster et al. 2010). Interestingly, the candidate phyla Poribacteria, which is highly abundant in various marine sponges over a wide geographic range (Fieseler et al. 2004), was recently described as part of the rare seawater biosphere (Webster et al. 2010). Webster et al. (2010) proposed that members of the rare seawater biosphere might act as seed organisms for widely occurring symbiont populations.

Sponge symbionts are capable of diverse metabolic functions. They play a crucial role in carbon, nitrogen, sulfur and phosphorous cycling and are also fundamental for the synthesis of essential vitamins within the sponge holobiont (reviewed by Webster and Thomas 2016). Examples of the mutualistic nature of sponge symbiont associations include the nitrifiying Thaumarchaeota and bacteria, which gain energy through the oxidation of ammonia, a sponge waste product (Webster et al. 2001; Bayer et al. 2008; Mohamed et al. 2010). Besides nutrient cycling capabilities within the host, symbionts are also shown to significantly contribute to host defense via the production of secondary metabolites (Hentschel et al. 2001; Kennedy et al. 2007).

Research on the resistance and resilience of the sponge microbiome to environmental stressors including local (e.g. sedimentation and nutrients) and global (e.g. elevated seawater temperature and ocean acidification) pressures has shown that responses are highly species-specific. For instance, a number of studies have revealed no compositional changes in the sponge-associated microbiome upon short-term nitrogen, temperature or sediment pulses, suggesting a highly resistant association between the sponge holobiont members (Luter et al. 2012, 2014; Simister et al. 2012). Similar observations were obtained during transplantation experiments of the sponge Aplysina cavernicola; light stressed sponges showed no changes in their bacterial community composition nor their production of secondary metabolites (Thoms et al. 2003). More recent research has shown that whilst the microbiome of different heterotrophic sponge species remains stable under light attenuation, the microbiome of phototrophic species can be significantly affected by light availability (Pineda et al. 2016). The symbiotic community associated with Rhopaloeides odorabile, a common Great Barrier Reef sponge, undergoes major changes in community structure, accompanied by host tissue necrosis, upon exposure to temperatures greater than 32 °C (Webster et al. 2008). A more recent metagenomic and metaproteomic analysis revealed that R. odorabile symbionts lose their metabolic functional potential during the early stages of heat stress and hence destabilize the sponge holobiont before visual signs of stress occur in the host animal (Fan et al. 2013). A comparison of healthy and diseased Ircinia fasticulata individuals revealed a significant shift in the microbiome prior to mass mortality attributed to high seawater temperatures. The observed shift was suggested to negatively affect host fitness and resistance to environmental stress (Blanquer et al. 2016). Numerous studies of sponge disease have reported a higher bacterial diversity in diseased tissue (Webster et al. 2008, Angermeier et al. 2011, 2012; Olson et al. 2014; Blanquer et al. 2016), suggesting that a dysbiotic microbiome rather than infection by a specific pathogen also has a role in the disease process. It is also noteworthy that some instances of ‘sponge disease’, do not involve detectable shifts in the composition of the microbial community (Luter et al. 2010), although no studies have yet assessed whether symbiotic function is impacted during disease which may contribute to the observed declines in host health.

Macroalgae and their microbiomes

Macroalgae have always been present in healthy coral reef ecosystems in relatively low abundance (Bruno et al. 2014). However, overfishing and eutrophication can facilitate a shift from coral dominated to algal dominated reefs (Hughes et al. 2007). The increase of macroalgal abundance on coral reefs is a threat for corals, not only because of direct competition for space but also because macroalgae have been shown to facilitate coral disease outbreaks, increase mortality and prevent larval settlement (Nugues et al. 2004; Smith et al. 2006; Sweet et al. 2013; Webster et al. 2015). Furthermore, macroalgae significantly alter their ambient environment by releasing higher amounts of labile dissolved organic carbon (DOC) compared to corals. This elevated DOC supports higher microbial growth rates, leads to a depletion of bioavailable DOC in seawater and facilitates the growth of copiotrophic and potentially pathogenic microbial taxa (Haas et al. 2011, 2016; Nelson et al. 2013). These processes have been summarized in the DDAM model (DOC, disease, algae, microorganism; Barott and Rohwer 2012), which describes the feedback of macroalgae-derived labile DOC and how it fuels the less efficient metabolism of copiotrophic and pathogenic microorganisms in coral reefs thereby contributing to increased microbial respiration and the local acidification of seawater (Sweet et al. 2013; Haas et al. 2016).

Besides their influence on the seawater microbiome, macroalgae themselves are holobionts and are associated with a highly diverse microbiota (Barott et al. 2011). Bacterial densities on the algal surface vary between 102 and 107 cells cm2 depending on the thallus section, host species and season (Armstrong et al. 2000). The epibacterial community composition significantly differs from the community in the ambient seawater and shows high host specificity and temporal adaptation (Lachnit et al. 2009, 2011; Burke et al. 2011b; Goecke et al. 2013). In addition to the biofilm community on the surface of seaweeds, macroalgae harbor a specialized and stable endophytic bacterial community (Hollants et al. 2011a, b). The specificity of epibacterial communities associated with different macroalgae is currently under reconsideration; microbial functioning, rather than phylogeny, seems to be consistent within the holobiont (Burke et al. 2011b, a).

The algal microbiome significantly contributes to host morphogenesis, health and defense. For example, morphological abnormalities were observed in various Ulva species when grown under aposymbiotic conditions (Provasoli and Pintner 1980; Nakanishi et al. 1996; Singh et al. 2011). Epiphytic bacteria, such as Vibrio sp. and Pseudoalteromonas sp., provide inhibitory properties against various biofouling organisms and hence contribute to host defense against unwanted colonization (reviewed by Egan et al. 2013). Additionally, epiphytic bacteria supply the algal host with key nutrients. Heterotrophic bacteria provide CO2 to the photoautotrophic host, and Cyanobacteria, dominant members of the epibacterial community, provide fixed-nitrogen (Penhale and Capone 1981; Phlips and Zeman 1990; de Oliveira et al. 2012). Recent genomic and metagenomic studies of seaweed-associated bacterial communities revealed a diverse genetic repertoire including genes for phosphorous, nitrogen and iron utilization (Thomas et al. 2008; Burke et al. 2011b; Fernandes et al. 2011).

In general, little is known about the factors controlling the microbial communities in the surface biofilm of macroalgae. Bacterial richness in the biofilm of the kelp Laminaria hyperborea is reported to increase as the kelp ages (Bengtsson et al. 2012) and microbial community succession seems to be influenced by stochastic processes (Burke et al. 2011a; Trias et al. 2012). Furthermore, disturbances such as temperature stress are shown to disrupt algal-holobiont homeostasis, which can lead to a switch in bacterial communities from surface-associated commensals to opportunistic pathogens (Case et al. 2011).

Methods to develop novel microbial indicators for coral ecosystem health assessment

High diversity, high functional complexity and low cultivability has historically limited our ability to understand the marine microbial realm. However, advances in culture-independent techniques along with next-generation sequencing (NGS) have revolutionized the study of microbial ecology (Schuster 2008). Accompanied by meta’omic approaches, NGS provides a practical tool for the efficient analysis of microbial communities in situ which will facilitate the identification of microbial indicators.

The first step towards identification of microbial indicators to assess coral reef health is analysis of the temporal and spatial variability of microbial communities associated with certain habitats and the subsequent definition of microbial baselines. Compositional baselines can be established with high-throughput sequencing of the 16S ribosomal RNA (rRNA) taxonomic marker gene of bacteria and/or archaea associated with environmental samples. This provides taxonomic insights and, when accompanied by functional prediction tools (e.g. PICRUST; Langille et al. 2013), may also help predict microbial functions within coral reef environments (Ainsworth et al. 2015; de Voogd et al. 2015). 16S rRNA gene sequencing is a ubiquitously applied technique in diverse fields of coral reef research (e.g. Webster et al. 2010; Bourne et al. 2013), providing an inexpensive tool to establish compositional baselines associated with coral reefs. In parallel, metagenomic sequencing can be used to establish functional and compositional microbial baselines as this comprehensive approach uses shotgun sequencing to generate an overview of all genes present in an environmental sample (including 16S rRNA genes). Metagenomics has been successfully used to assess functional responses of the coral microbiome to various stressors (Vega Thurber et al. 2009; Littman et al. 2011; Kelly et al. 2014) and is also frequently applied to understand the fundamental basis of the sponge (Thomas et al. 2010; Fan et al. 2013), macroalgae (Burke et al. 2011a; Martin et al. 2014) and planktonic (Dinsdale et al. 2008; Tout et al. 2014; Haas et al. 2016) microbiomes.

Once baselines are established, the compositional and functional response of the microbial communities upon disturbance can be investigated. Taxonomic variation can be observed using 16S rRNA gene sequencing and compositional and functional shifts can be detected using metagenomic sequencing. However, metagenomics is limited in its ability to describe functional variations upon disturbance because of the natural discrepancy between genes that are present and genes that are actually being expressed (Wang et al. 2015). Metatranscriptomic sequencing overcomes this limitation by assessing the function encoded by mRNA sequences. Metatranscriptomics involves analyzing near-real-time gene expression patterns by sequencing mRNA molecules and has emerged as a state-of-the-art tool to study community metabolism of free-living microbes in the open ocean (Poretsky et al. 2009; Hewson et al. 2010) and to obtain insights in cell–cell signaling, development and immune response of symbiotic interactions (Chun et al. 2008; Ruby 2008; Stewart et al. 2011; Sanders et al. 2013). It has also been used to elucidate nutrient cycling and vitamin production pathways of a sponge holobiont (Fiore et al. 2015). Furthermore, responses of the coral host and its associated microbiome to coral diseases have been investigated based on mRNA sequences (Daniels et al. 2015; Arotsker et al. 2016). However, a clear limitation of this method is the short lifetime of mRNA molecules; it only takes a few minutes before mRNA molecules are degraded within the cell (Pedersen et al. 2011). Metaproteomics can characterize the protein signatures from microbial communities in situ and also provides a link between gene content and gene expression (von Bergen et al. 2013). Thus, metaproteomic studies are often complemented by metagenomic data. For example, the combination of metagenomic and metaproteomic data on the sponge microbiome provided novel insights in the activity, physiology and interactions between sponge symbionts (Liu et al. 2012) and revealed the functional role of microbes in the stability of the sponge holobiont under thermal stress (Fan et al. 2013).

Finally, a quantitative approach that establishes links between the composition / function of reef microbiomes and environmental metadata will be required to identify microbial indicators for coral reef health and water quality. Individual microbial taxa and/or functions significantly associated with healthy versus stressed reef systems can be identified using traditional statistical approaches such as the IndVal analysis (Dufrêne and Legendre 1997). An alternative approach to analyzing meta’omics data and discriminating between healthy and stressed reef ecosystems is machine learning. Instead of the identification of individual microbial indicators, the entire microbial community and its function could be used to train a model to differentiate between reef health stages or environmental stressors. Machine learning is a powerful tool and a current state-of-the-art approach to identify dysbiosis of the human microbiome and to predict human diseases (Pasolli et al. 2016).

Conclusion and future research

Microorganisms are fundamental contributors to reef ecosystem health through their biogeochemical capabilities and intimate symbiotic partnerships. Shifts in the composition or function of bacterial and archaeal communities can therefore provide crucial diagnostic information for future coral reef monitoring. Before such approaches can be developed and implemented, the following basic questions need to be resolved (see Fig. 2): (1) which taxa and functions form the microbial baseline of healthy coral reefs? (2) how does the microbial community respond to environmental change? (3) how does the microbiome influence holobiont resistance and resilience upon disturbance? and (4) are environmental disturbances predicted by compositional and functional changes in the microbial community?

Fig. 2
figure 2

Schematic representation of the outlined future research directions: The microbial baseline (horizontal bar) needs to be defined for a comprehensive range of coral reef habitats (including seawater, sediment, sponges, corals and macroalgae), over temporal periods for a particular microbial community parameter of interest (e.g. diversity). A disturbance event (vertical bar) alters the environmental conditions, and triggers a shift in the microbial community associated with each habitat (solid black line) leading to an alternative stable state that may significantly deviate from the natural state. The stability of a holobiont, which relates to its ability to act as a buffer to maintain coral reef function upon environmental alteration, is here defined by the combined resistance and resilience of the microbial community. Resistance is the ability of a community to withstand a disturbance without change. Resilience is defined as the rate at which a microbial community regains its original state after a disturbance event. Combining the information on microbial community stability and variation will enable identification of microbial indicators leading to the early identification of imminent environmental stressors (e.g. water quality changes)

Regular monitoring is a fundamental tool for conservation and resource management of marine ecosystems in both developed and developing countries. For example, in the Australian GBR extensive reef monitoring records coral abundance, disease prevalence and coral cover, with the overarching objective of determining the status of reef health and pinpointing changes in the distribution and abundance of the reef biota over large temporal and spatial scales (AIMS 2017). Extending already existing monitoring initiatives to include sample collections targeting microbial biodiversity, composition and function would provide a cost-effective strategy to establish the first microbial reference datasets for individual reef locations (Phase 1). By combining microbial community data and other environmental parameters (e.g. water quality), microbial indicators (taxa or functions) can be identified allowing for reef health diagnosis (Phase 2). Once microbial indicators have been determined, the development and testing of cost- and time-efficient microbial monitoring protocols can begin (Phase 3). The ultimate goal of Phase 3 is the establishment of reliable, fast, low-cost and easy-to-use diagnostic protocols based on microbial indicators that can be integrated into current monitoring programs. Microbial monitoring protocols can be comprised of targeted PCR-based approaches (e.g. PCR screening for the occurrence of a specific microbial taxon/function, or quantification of its abundance) and/or community sequencing approaches (e.g. amplicon sequencing to track shifts in microbial community composition). The final step (Phase 4) is the integration of microbial indicators into standard reef monitoring procedures. A prospective microbial indicator tool kit will offer streamlined procedures covering sample collection and processing, an online data analysis platform, and recommended guidelines for management interventions based on the diagnosed reef conditions.