Disturbances of coral reefs are caused by a complex combination of stressors including those arising from climate change, diseases, predation, destructive fishing practices, storms and changes in water quality. Many of these stressors are increasingly acting over regional and/or global scales; therefore, maintaining the resilience of coral communities will require a reduction in local, but manageable, impacts such as altered water quality. However, in some parts of the world, nutrient and sediment inputs to coral reefs have increased several-fold over the last 150 years (e.g. Richmond 1993; McCulloch et al. 2003) causing reduced coral recruitment (Loya 1976; Babcock and Davies 1991; Loya et al. 2004), modified trophic structures (Lapointe 1997; Fabricius 2005), altered biodiversity (van Woesik et al. 1999) and coral mortality (Kline et al. 2006). These processes can transfer the competitive advantage away from reef-building corals leading to trophic dominance by assemblages of macroalgae (Schaffelke 1999) once productivity exceeds rates of herbivory (McCook 1999). Therefore, monitoring programmes require a suite of bioindicators that can effectively quantify the link between changes in water quality and the condition of coral-reef ecosystems.

The use of bioindicators provides a number of significant advantages over direct measurements of water quality. For example, a direct measurement of water quality provides information about the condition of the water column at that particular point in time. Moreover, if the frequency of sampling is limited, or is weather-dependant and constrained by safety considerations, then important information on the effects of acute episodic events that can strongly influence the structure of coral communities may not be quantified (e.g. terrestrial discharges during floods or the resuspension of sediments during strong winds). These issues are addressed with the use of appropriate bioindicators that provide a time-integrated measure (from time periods of minutes to years) of the effects of changes in water quality on coral reefs. Given the range of natural and anthropogenic factors that influence a complex ecosystem such as a coral reef, compared with a single bioindicator a suite of bioindicators of cellular, organismal and community effects will more effectively attribute ecological change to changes in specific environmental conditions (Erdmann and Caldwell 1997; Jameson et al. 1998). Suites of bioindicators and predictive models have been developed and applied successfully for assessments of ecosystem health in estuarine and freshwater systems. For example, models such as AUSRIVAS (Simpson and Norris 2000) assess ecosystem health based on assemblages of freshwater macroinvertebrates, and the SIGNAL biotic index uses the presence or absence of families of macroinvertebrates to infer levels of exposure of river systems to chemical pollutants (Chessman et al. 1997). Few studies have assessed the suitability of potential bioindicators or their application to an index of coral-reef ecosystem health (Brown 1997). Three notable exceptions exist. First, Risk et al. (2001) used the diversity of certain invertebrates (e.g. stomatopods and amphipods), measures of bioerosion and geochemical markers to quantify the health of coral-reef ecosystems exposed to terrestrial runoff. Second, Jameson et al. (2001) developed Indexes of Biotic Integrity (IBI’s) for coral reefs combining several metrics for sessile epibenthos, benthic macroinvertebrates, fish, marine vegetation, phytoplankton and zooplankton. Third, Jameson and Kelty (2004) reviewed potential bioindicators of changes in sediments, nutrients, heavy metals, herbicides, pesticides and bacterial exposure, as well as multi-metric bioindicators of the combination of stressors on scleractinian and non-scleractinian communities.

This review builds upon these previous studies by further investigating the desirable characteristics of a suite of bioindicators and their response time to changes in water quality. Specifically, the aim of this review was to assess and identify the bioindicators that are potentially most appropriate for use in monitoring programmes that link changes in water quality (i.e. dissolved and/or particulate nutrients, rate of sedimentation, turbidity, benthic irradiance, or other pollutants such as herbicides and pesticides) to changes in the condition of coral-reef ecosystems. Candidate bioindicators were first grouped according to the ecological level at which the response to changes in water quality could be measured, i.e. genetic/colony, population or community level. This initial grouping also corresponded to differences in response times, with bioindicators at the genetic/colony level characterised by rapid responses to changes in water quality, whilst responses at higher levels of ecological organisation generally occurred over longer time periods. This has important implications for the selection of bioindicators because the timeframe available for monitoring can vary from years or decades for some long-term monitoring programmes, and weeks or months for short-term monitoring programmes such as environmental impact assessments (EIA). Within this context, the scientific literature was used to rank the suitability of bioindicators according to a framework that assessed whether responses to changes in water quality were (i) specific, (ii) monotonic, (iii) variable, (iv) practical and (v) relevant to management goals for long- and short-term monitoring programmes.

Desirable characteristics and selection of bioindicators

A change in water quality may be chronic (e.g. altered runoff regime) or acute (e.g. dredging or flood events), so an important characteristic in the selection of any bioindicator is the time taken for the biological response to manifest at the genetic/colony, population or community level. Both the duration of response initiation and the recovery period can range from near-instantaneous to decades. An important distinction, therefore, is to differentiate between measures suitable for detecting effects during or shortly after exposure to a change in water quality (i.e. rapidly, whereby the response initiation and recovery from an event occurs within hours to weeks), and those better suited to detecting cumulative effects over prolonged periods of time (i.e. slowly, whereby initiation of and recovery from response may take months to years). There are both advantages and disadvantages to rapid and slow response times in bioindicators according to their application. A rapid response could be considered desirable as an ‘early warning’ indicator of change, i.e. a sublethal response used in short-term monitoring programmes that incorporate reactive management strategies to mitigate the effects of any further changes in water quality. This is offset, however, by the high level of sampling intensity and replication required to obtain accurate estimates of a response that could change over days to weeks. Conversely, a bioindicator responding over a longer timescale may not provide an early warning of change, but be particularly useful for monitoring chronic exposures, as these types of bioindicators are likely to have low-natural variability and require fewer samples to detect ecological change. Most importantly, the response time of a bioindicator to a change in water quality must be comparable to the timeframe of the disturbance being measured.

In order to select a bioindicator objectively requires a set of selection criteria. Here, five key criteria were used to rank the suitability of bioindicators for assessing changes in water quality on coral reefs (Table 1; modified from Jones and Kaly 1996; Erdmann and Caldwell 1997; Jameson et al. 1998). First, specificity is the extent to which the changes in the bioindicator respond to changes in water quality, and not to other environmental conditions (Fig. 1a). Second, monotonicity is the extent to which the magnitude of the changes in the bioindicator are proportional to the intensity and duration of the changes in water quality, which is evident in the shape of the dose–response relationship (Fig. 1b, c). Third, variability is the extent of natural variation of the bioindicator in the absence of changes in water quality. A bioindicator that displays patterns of seasonal or temporal variability might still be suitable for monitoring programmes provided that the effect size at the time and location of the disturbance differs significantly compared with variability at reference locations (Fig. 1d). Fourth, practicality is the extent to which the changes in the bioindicator are easily quantified, and depends on cost, observer independence, level of expertise and the spatial and temporal scales required for application. Last, relevance refers to ecological relevance as well as to relevance in public perception, which will assist in the communication of the results to a wide range of end-users.

Table 1 Criteria for selection of bioindicators to assess effects of changes in water quality on corals and coral communities
Fig. 1
figure 1

Response of a hypothetical bioindicator to a disturbance (dark grey areas represent the disturbance in question; light grey areas represent other disturbances). A suitable bioindicator must detect differences between a disturbed state (solid line) and reference states (dashed lines). a response of a bioindicator with high specificity, b and c monotonic dose-response relationships whereby the magnitude of change of the bioindicator is proportional to the intensity and duration of the disturbance, d responses at impact locations must differ from those at reference locations if bioindicators that vary temporally are used in monitoring programmes

Bioindicators of the effects of changes in water quality on coral reefs

From the many bioindicators presented in the literature, 21 candidate bioindicators were short-listed for potential use in monitoring programmes on coral reefs. The next section and Table 2 provide a brief description of the main properties of these bioindicators. In order to facilitate the systematic comparison of bioindicators and to assess their suitability to detect either chronic or acute changes in water quality, an assessment framework (Table 3) was developed based on the five selection criteria. Each of the candidate bioindicators was scored against each criterion, and the sum of positive scores determined their final rank at a scale of 1–5, with each criterion having the same weight. Hence, a bioindicator would receive a maximum score of 5 if it responds specifically and monotonically to changing water quality, has low background variability, is practical to implement and has high ecological and public relevance. Bioindicators with a score of 4 or 5 were assigned a ‘high-priority’ ranking for use in either long- or short-term monitoring programmes, depending on their response time. Bioindicators that ranked 2–3 were given a ‘medium priority’ due to satisfying only some of the selection criteria, while a ranking of <2 may provide useful, often complimentary, information about the responses of corals to changes in water quality, but were considered ‘low-priority’.

Table 2 Examples of studies examining coral reef bioindicators to changing water quality at the level of individual colonies, populations and communities on coral reefs
Table 3 Assessment framework for identifying bioindicators of the effects of changes in water quality on coral reefs

Genetic/colony measures

Gene expression

Genetic biomarkers are emerging as powerful tools to identify sources of stress and measure stress responses in corals (Edge et al. 2005; Morgan et al. 2005) and their symbionts (Leggat et al. 2007 and see reviews by van Oppen and Gates 2006; Foret et al. 2007). Regulation of stress-specific genes is determined by comparing gene expression in populations exposed to environmental stressors with those at reference conditions. For example, the development of a complimentary DNA (cDNA) array containing 32 stress genes allowed the profiling of gene expression in corals exposed to changes in seawater temperature, salinity and ultraviolet light in the laboratory (Edge et al. 2005) and with increasing distance from leachate associated with a municipal dump (Morgan et al. 2005; Table 2). Other studies have investigated changes in the expression of large numbers of genes with DNA microarrays for stressors, such as changing seawater temperatures (Foret et al. 2007), suggesting potential for similar applications to detect changes in water quality. Gene expression ranked a medium-priority bioindicator for use in long- and short-term monitoring programmes (Table 3). Whilst it has been shown to be highly specific, the methods to detect changes in water quality require testing under field conditions and the patterns of temporal variability are not well understood.

RNA/DNA ratio

The RNA/DNA ratio is based on the principle that the amount of DNA within a cell remains invariant whereas the amount of RNA varies as a function of metabolic processes, such as protein synthesis related to growth. Under conditions favourable for coral growth, such as in clear-water with high benthic irradiance, the RNA/DNA ratio is expected to be greater than conditions where growth is limited by available light. Indeed, recent studies have demonstrated positive relationships between RNA/DNA ratio and light intensity along depth and turbidity gradients (Meesters et al. 2002; Buckley and Szmant 2004; Table 2). However, interspecific and seasonal variability, coupled with complex responses to increased heterotrophy in turbid conditions, suggests that although RNA/DNA ratios may be sensitive to changes in water quality, further research is required to better understand the sources of variability. The RNA/DNA ratio ranked a medium-priority bioindicator for use in long- and short-term monitoring programmes (Table 3), until a better understanding of interspecific and seasonal variability in this ratio is obtained. This bioindicator may, however, prove useful in EIA to measure effects of acute changes in water quality within species when values from impact locations are compared with those from reference locations.

Symbiont photophysiology

The photophysiology of coral symbionts can be characterised by several parameters each having different responses depending on exposure to the various components that comprise a change in water quality. For example, the maximum quantum yield (F v /F m) decreases with increasing exposure (duration and quantity) to sediment (Philipp and Fabricius 2003) and exposure to agricultural chemicals (Jones et al. 2003; Table 2), but appears not be influenced (for cultured symbionts) by exposure to elevated levels of dissolved inorganic nutrients (Rodriguez-Roman and Iglesias-Prieto 2005). Exposing the symbionts to a series of irradiances in short (10 s)-incremental steps producing a rapid light curve can also provide detailed information on photo-acclimatory responses of corals to changes in water quality (Ralph and Gademann 2005). Quantitative parameters derived from rapid light curves include maximum photosynthetic rate (PSmax), minimum saturating irradiance (E k) and light utilisation coefficients (α) of the initial rise of the curve. Symbionts acclimatised to high irradiances are characterised by high PSmax and E k, with low α, but the opposite occurs for symbionts acclimatised to low-irradiance as the symbionts attempt to optimise their light capture and utilisation capability (Anthony and Hoegh-Guldberg 2003; Cooper and Ulstrup 2009). However, these parameters are influenced by variation in seawater temperature (Coles and Jokiel 1977; Warner et al. 1996; Fitt et al. 2001), flow regime (Nakamura et al. 2005), diurnal changes in benthic irradiance (Jones and Hoegh-Guldberg 2001; Lesser and Gorbunov 2001) and symbiont genotype (Frade et al. 2008; Hennige et al. 2009). Thus, these factors must be accounted for when using symbiont photophysiology to infer changes in water quality. Nevertheless, photophysiological responses are rapid (i.e. timescales of minutes to days) making these measures particularly suitable as sublethal bioindicators of changes in water quality if confounding variables are controlled. Symbiont photophysiology is, therefore, ranked a high-priority bioindicator for use in long- and short-term monitoring programmes (Table 3). Chlorophyll a fluorometers can be used to quantify photophysiological responses although the application of these instruments as monitoring tools requires a high level of training and a theoretical understanding of their utility.

Colony brightness

The colour of scleractinian corals is determined by photosynthetic pigments contained in the algal endosymbionts and light-absorbing compounds in the coral tissue (e.g. Jeffrey and Haxo 1968; Salih et al. 2000; Dove et al. 2001), both of which are known to respond to changes in water quality. Concentrations of chlorophyll a (and hence colour brightness) increase in response to exposure to elevated nutrients (Hoegh-Guldberg and Smith 1989; Table 2) and reduced irradiance (Falkowski and Dubinsky 1981; Dubinsky et al. 1984) whereas symbiont density may decrease in response to sedimentation (Nugues and Roberts 2003) and exposure to pollutants, such as cyanide (Cervino et al. 2003). However, symbiont density also varies with season (Stimson 1997; Brown et al. 1999; Fagoonee et al. 1999) and seawater temperature, indicating a moderate specificity to changes in water quality. Correlating colony brightness to changes in water quality also requires large spatial and temporal replication in monitoring programmes because photo-acclimatory responses occur on short timescales (Anthony and Hoegh-Guldberg 2003). Changes in host pigmentation in response to changes in water quality are less well documented, although Salih et al. (2006) found a modal response in levels of fluorescent pigments across a depth gradient and suggested the possibility of a dual role for these host pigments; photo-protective at high irradiances and light-amplifying at low irradiance deep depths. Notwithstanding these caveats, colony brightness ranked a high-priority bioindicator for use in long- and short-term monitoring programmes (Table 3) because colour changes can be measured with simple tools, such as colour charts (Siebeck et al. 2006), making it a useful sublethal bioindicator and a trigger for more intensive studies.

Lipid content

Corals are mixotrophic organisms that assimilate carbon from their symbionts (Muscatine 1980), the capture of zooplankton (Porter 1974) and the digestion of organic particulate matter (Tomascik and Sander 1985; Anthony 1999). Any carbon that is surplus to metabolic requirements is excreted, or stored as energy reserves in the form of lipids (Crossland 1987; Anthony and Fabricius 2000). Lipid content can change with exposure to dissolved nutrients (Achituv et al. 1994), altered light availability (Stimson 1987; Harland et al. 1992) and turbidity (Anthony and Fabricius 2000; Table 2). Saunders et al. (2005) linked differences in the ratio of non-polar storage lipids to polar structural lipids within corals, to acute changes in water quality by a relative decrease in non-polar storage lipids; the stress response was a net decrease in the amount of energy available for storage as lipids. Importantly, the use of such a ratio removes the need to normalise data, which along with potential loss of lipid mass during the gravimetric analysis, introduces another source of uncertainty into the estimate of total lipid content. Moreover, total lipid content may vary according to the production of mucus and gametes (Stimson 1987), seasons (Oku et al. 2003), and is lowest at times of high metabolic demand such as during periods of rapid growth or reproduction (Leuzinger et al. 2003). Nevertheless, this bioindicator has a rapid response time with changes occurring within weeks following changes in water quality (Anthony and Fabricius 2000; Saunders et al. 2005). Determination of total lipid content ranked a medium-priority bioindicator for use in long- and short-term monitoring programmes (Table 3) because of the high level of intra-colonial and seasonal variability associated with this measure. However, relative differences in the ratio of non-polar/polar lipids of corals show great potential for use in EIA when comparing impact with reference locations. Analysis of lipid content requires small samples to be collected from the coral colony, followed by analytical techniques such as gravimetry for total lipid content, or using thin-layer chromatography for determination of lipid fraction ratios (Saunders et al. 2005).

Tissue thickness of massive corals

Tissues of massive Porites have been found to be thicker on coastal reefs compared with offshore reefs, possibly because of the higher concentrations of nutrients and particulate organic matter in the coastal zone (Barnes and Lough 1992; Table 2). Many environmental factors, including irradiance, seawater temperature, water clarity, particulate and dissolved inorganic nutrient levels and sedimentation change from inshore to offshore locations, and previous studies have not resolved which of these factors is the most influential in affecting tissue thickness. Barnes and Lough (1992) suggested that massive Porites would have thicker tissue layers where particulate matter and other food items were not limited. Tissue thickness can change within a few weeks following a change in water quality (Cooper 2008), and hence this measure ranked a high-priority bioindicator for use in long- and short-term monitoring programmes (Table 3). Tissue thickness is best determined in massive Porites by removing a small core from the upward-facing surface and measuring the depth of the skeleton occupied by living tissue with callipers. However, as the thickness of the tissue layer can vary within a colony (Barnes and Lough 1992), sampling must be standardised and replicated. A limitation of this measure is that the sampling is intrusive, and so procedures that mitigate sampling effects, i.e. plugging core-holes to facilitate tissue regrowth, need to be considered.

Surface rugosity of massive corals

The surface rugosity of massive Porites increases when skeletal growth is unable to provide sufficient surface area to accommodate increased tissue growth. Several studies have shown that Porites colonies had a greater surface rugosity on coastal than offshore reefs of the Great Barrier Reef, potentially due to increased availability of nutrients and particulate organic matter (Darke 1991; Scoffin et al. 1992, Table 2), suggesting a high specificity to changes in water quality. Thus, surface rugosity ranked a high-priority bioindicator for use in long-term monitoring programmes (Table 3). It is, however, likely to be of limited use for short-term monitoring due to its slow response time. Surface rugosity can be determined by placing a piece of chain of known length on the upper surface of the colony and calculating the ratio between the horizontal and vertical chain length (Darke 1991).

Coral growth

Irradiance is a key determinant of coral growth (Goreau and Goreau 1959; Bak 1974). However, coral growth is also indirectly affected by changes in irradiance due to variable turbidity and nutrient conditions (e.g. Marubini and Davies 1996). On the Great Barrier Reef, the growth parameters of massive Porites vary along cross-shelf gradients. For example, the skeletal density of massive Porites increases (Risk and Sammarco 1991), but the rate of linear extension and calcification decreases, with increasing distance from the coast (Lough and Barnes 1992; Table 2). However, coral growth is also influenced by variation in other parameters such as temperature, thermal stress and ocean acidification (Lough and Barnes 2000; Cooper et al. 2008a; De’ath et al. 2009) indicating a moderate specificity to changes in water quality. Coral growth ranked a medium-priority bioindicator for use in long-term monitoring, but was of low priority for short-term monitoring because of its slow response time (Table 3). Notwithstanding this, analysis of coral growth records provide important historical information on environmental changes over decades to centuries that can be used in combination with other bioindicators in long-term monitoring programmes. The coral growth parameters are generally derived from cores or slices of massive corals (Knutson et al. 1972), but can also be measured in corals with other growth forms (Ferrier-Pages et al. 2000; Table 2). Techniques for quantifying coral growth include gamma densitometry (Chalker and Barnes 1990) to determine skeletal density, with the distance between the peaks of adjacent density bands to determine linear extension, and the product of these two parameters, linear extension and density, to estimate calcification rate.

Skeletal elemental and isotopic composition

Chemical elements incorporated from the water column into the coral skeleton can be used as retrospective bioindicators of environmental conditions (e.g. Goreau 1977; Risk et al. 2001; Cohen and McConnaughey 2003; Cohen et al. 2004). Several studies have examined the utility of the ratio of the nitrogen isotopes 15N/14N (known as δ15N) to hindcast changes in water quality on coral reefs (Sammarco et al. 1999; Heikoop et al. 2000; Table 2). The main sources of anthropogenic nitrogen (sewage, organic nitrogen from soil erosion and synthetic fertilisers) each produce characteristic δ15N signatures that allow identification of the source of nitrogen (Heaton 1986). Sewage effluent is enriched with the heavier 15N isotope and has δ15N values in the range from +10‰ to +22‰, soil-organic nitrogen between +4 and +9 ‰, with synthetic fertilizers from –4‰ to +4 ‰ (Heaton 1986). For example, Marion et al. (2005) traced the introduction of synthetic fertiliser on rice fields in Bali back to the early 1970’s using δ15N records in coastal corals exposed to agricultural runoff. Skeletal elemental and isotopic composition is, therefore, ranked a high-priority bioindicator for use in long- and short-term monitoring programmes (Table 3). Analyses of skeletal chemistry typically involve the collection of a small sample of coral followed by analysis using mass spectrometry to identify specific sources of pollutants on coral reefs (Risk et al. 2001).

Partial mortality

A partial mortality is a lesion in the living tissue of a coral colony that may be caused by changes in a range of environmental conditions (Hughes and Jackson 1980). Estimates of partial mortality on the surfaces of coral colonies have been used to assess the effects of water quality stressors such as sedimentation on coral reefs. Nugues and Roberts (2003) reported higher partial mortality in some coral species at sites closest to rivers compared with those at some distance from rivers (Table 2). Importantly, this response was species specific and likely to be greater in species with poor sediment-rejection abilities (Obura 2001; Nugues and Roberts 2003). Ginsburg et al. (2001), however, found great variability in the partial mortality of massive corals and suggested that for corals with a massive growth form, partial mortality was not a suitable bioindicator of changes in water quality along the Florida Reef Tract. Partial mortality can also result from injuries due to competitive interactions with other benthic organisms, feeding scars from coral predators, overgrowth by macroalgae, coral diseases, storms or other physical damage (Garzon-Ferreira et al. 2005) indicating low specificity to changes in water quality. Thus, partial mortality ranked a medium-priority bioindicator for use in long-term monitoring programmes (Table 3). Increases in rates of partial mortality are, however, likely to be rapid following acute changes in water quality such as increased sedimentation. Thus, measuring rates of change in partial mortality ranked a high-priority bioindicator for short-term monitoring and could be used as a trigger for a more intensive impact assessment. Partial mortality can be quantified by estimating the proportion of colony surface free of living tissue, or using photographic techniques to measure the area of lesions and colony surfaces.

Mucus production

Mucus is produced by some corals to clean the colony surface of sediments, or as a response to emersion during low tides, turbidity, pollutant exposure, changes in salinity and water temperatures and injury (Stafford-Smith and Ormond 1992; Stafford-Smith 1993; Brown and Bythell 2005). Some corals, however, will not produce mucus when stressed, or will stop producing mucus after prolonged stress. For example, sheets of mucus may only be produced by massive Porites to remove sediment when conditions are calm (Coffroth 1985), and fungid corals may cease producing mucus when cells become ‘exhausted’, despite ongoing exposure to stressors (Schuhmacher 1977). Mucus production by corals ranked a low priority bioindicator for use in long- and short-term monitoring programmes (Table 3). Mucus production is difficult to quantify and currently best recorded as qualitative observations. However, excess mucus production may be a useful qualitative bioindicator for short-term monitoring provided that complimentary evidence is obtained from quantitative methods.

Population measures

Population structure

The structure of a population can be defined as the number of individuals at different life-history stages. The effects of changes in water quality on coral reefs have been inferred by quantifying changes in the structure of coral populations through time (e.g. Bak and Meesters 1999; Meesters et al. 2001; Gilmour 2004). Meesters et al. (2001) found that the size–frequency distributions of coral populations closest to urban centres were negatively skewed, having fewer juveniles and more large colonies than reference locations. For some species, however, differences in size–frequency distributions were related to their life-history strategies, such as the degree to which corals invest in different modes of sexual and asexual reproduction (Gilmour 2004), or species with small, short-lived colonies (Meesters et al. 2001). In contrast, the size–frequency distributions of populations recovering from a bleaching event were skewed positively (dominated by juvenile size classes) at a location exposed to infrequent floods, whereas it was platykurtotic (flattened) at a location with a higher flood frequency (Smith et al. 2005; Table 2). Population size–frequency distributions of corals are affected by a range of other confounding variables such as flow, irradiance, connectivity to larval sources and disturbance history (Hughes 1989; Hughes and Connell 1999), indicating a low specificity to changes in water quality. Thus, population structure ranked a medium-priority bioindicator for use in long- and short-term monitoring programmes (Table 3). The population structure of a coral reef can be assessed by measuring sizes and determining the abundance of colonies along transects.

Coral diseases

Coral diseases are recognised as a major form of disturbance for coral reefs (Sutherland et al. 2004) and anthropogenic influences are considered to increase their prevalence (Green and Bruckner 2000; Bruno et al. 2003). At present, approximately 20 diseases have been identified that affect more than 100 species of corals (Sutherland et al. 2004). Most diseases are caused by pathogens such as bacteria, cyanobacteria and fungi. Pathogens can be delivered to coral reefs via terrestrial runoff (Sutherland et al. 2004), or transmitted by biological vectors (Sussman et al. 2003). Recent studies have highlighted the role of elevated levels of dissolved organic carbon, a common component of sewage and organic discharge, to tissue sloughing (symptomatic of many diseases) and increased coral mortality (Table 2; Kuntz et al. 2005; Kline et al. 2006). However, elevated water temperatures can also increase the spread and virulence of pathogens on reefs with high coral cover (Bruno et al. 2007) and reduce the immune response in corals, indicating a moderate specificity to changes in water quality. The incidence of coral disease ranked a medium-priority bioindicator for use in long-term monitoring programmes (Table 3). Notwithstanding this, relative changes in the incidence of coral diseases may provide useful information for control-impact type studies. Coral disease prevalence can be quantified as the proportion of colonies with visible disease symptoms in large quadrats or along transects (e.g. Cervino et al. 2001).

Abundance of macro-bioeroders

Bioerosion is the process of erosion of substrata by biological activity, and comprises both internal bioerosion (i.e. micro- and macroborers) and external bioerosion (i.e. grazers) (Hutchings 1986; Bellwood 1995). Rose and Risk (1985) found an increased abundance of the boring sponge Cliona delitrix on reefs associated with the discharge of untreated sewage. Similarly, the inverse relationship between the abundance of internal bioeroders and distance from the coast, in Acropora formosa and massive colonies of Porites, was attributed primarily to a greater exposure to terrestrially derived nutrients on nearshore reefs on the Great Barrier Reef (Sammarco and Risk 1990; Risk et al. 1995; Table 2). Abundances of macro-bioeroders have been shown to have a high specificity and slow-response period to spatial differences in water quality (Cooper et al. 2008b), and hence provide a useful time-integrated measure of water-quality changes. For this reason, abundance of macro-bioeroders ranked a high-priority bioindicator for use in long-term monitoring, but had medium priority for short-term monitoring because of the slow-response period (Table 3). The abundance of macro-bioeroders (predominantly sponges, polychaetes, sipunculans, barnacles and bivalves) can be determined with counts of external bore holes occurring in quadrats placed on the surface of living massive corals.

Community measures

Micro- and meiobenthic bioindicators

A variety of micro- and meiobenthic measures are potential bioindicators of changes in water quality. For example, the content of organic matter (Hyland et al. 2005), abundance and composition of bacteria species (Uthicke and McGuire 2007) and microphytobenthos (Bell and Elmetri 1995; Gottschalk et al. 2007) of sediments, as well as foraminifera (Cockey et al. 1996; Uthicke and Nobes 2008), amphipods (Thomas 1993) and stomatopods (Risk et al. 2001) are all potential bioindicators of changes in water quality. Indeed, some of these indicators have already been applied successfully in monitoring programmes e.g. the FORAM Index developed by Hallock et al. (2003). Whilst most micro- and meiobenthic indicators show great potential for application to coral reefs, their widespread acceptance will depend on the availability of taxonomic expertise or genetic assays, or the development of simple protocols that overcome the need for species differentiation. Recently, Uthicke and Nobes (2008) demonstrated the application of a FORAM Index to coral reefs by quantifying the disappearance of large symbiont-bearing taxa in foraminifera assemblages along a gradient of decreasing irradiance and increasing nutrient availability, from the outer to the inner Great Barrier Reef (Table 2). Micro- or meiobenthic bioindicators, especially foraminifera, were ranked high-priority bioindicators for use in long- and short-term monitoring programmes (Table 3). Methods such as the FORAM Index are based on easily quantifiable size classes, rather than species composition, thereby useful information can be obtained even by observers with limited taxonomic expertise (Hallock et al. 2003). Foraminifera communities can be quantified simply by collecting replicate sediment samples on coral reefs, which may be dried and stored for later processing.

Larval supply and coral recruitment

Many spawning corals release gametes synchronously over a few nights each year that develop into larvae and disperse for days to months before settling and metamorphosing into coral polyps. The number of larvae settling at a site is a measure of larval supply, while the number of small juvenile corals (typically defined as <5 cm) is a measure of recruitment to the community. High sedimentation, turbidity and nutrients all reduce the number of larvae produced by corals, their rates of settlement and early post-settlement survival (Babcock and Davies 1991; Hunte and Wittenberg 1992; Wittenberg and Hunte 1992). Larval supply of corals is influenced by the abundance and fecundity of adult corals on the reef and on adjacent reefs, the abundance of predators and competitors, and the physical conditions before, during and after spawning (e.g. Hughes and Jackson 1985; Hughes and Connell 1987). Consequently, there is great natural variation in the supply of larva on a coral reef. Thus, larval supply ranked a low priority bioindicator for use in monitoring programmes (Table 3). In contrast, larval settlement is typically the greatest on surfaces that are relatively free of sediment, i.e. vertical or downward-facing horizontal surfaces in areas of high sedimentation (Babcock and Davies 1991; Gilmour 1999; Table 2). Thus, the ratio of coral recruits on vertical, compared with upward-facing horizontal, surfaces may prove a better bioindicator of changes in sedimentation on coral reefs than settlement abundance alone, but field comparisons are required to determine suitability. Coral recruitment (i.e. the number of small juvenile corals of a given taxonomic group) ranked a high-priority bioindicator for use in long- and short-term monitoring programmes, because changes in the abundance of the smallest colony size classes reflect their susceptibility to changes in water quality such as rates of sedimentation (Table 3). Larval supply can be quantified using larvae traps (Lasker et al. 1998) or settlement plates (e.g. Babcock and Davies 1991), and changes in the number of coral recruits can be quantified within quadrats or transects, although both methods are labour intensive and require some taxonomic expertise.

Benthic cover

Benthic cover of coral-reef biota is perhaps the most widely measured parameter in coral-reef monitoring programmes (e.g. Marshall and Orr 1931; Smith et al. 1981; Bell and Elmetri 1995). Estimates of the percentage cover of benthic communities are generally calculated for hard corals, octocorals, macroalgae and sediment, and the proportion of these categories has been shown to vary with changes in water quality. For example, coral cover is known to decrease with declining water quality in the Indo-Pacific (e.g. Brown et al. 1990; van Woesik et al. 1999; Table 2). In contrast, Lirman and Fong (2007) found a negative relationship between coral cover and changes in water quality in the Florida Keys, where coral cover was significantly greater on coastal compared with offshore reefs, despite the existence of a water quality gradient. In some instances, coral cover alone may not reflect the effects of changes in water quality, because coral cover is also influenced by other forms of disturbance, which include cyclones, destructive fishing, bleaching events, predation by Acanthaster planci and ship groundings. Therefore, changes in coral cover have a low specificity to changes in water quality, and using it as a co-variate in water-quality assessments will require knowledge of the disturbance history. Response times of loss of coral cover vary from weeks for acute changes in water quality (Brown et al. 1990), to years for chronic exposure. Thus, benthic cover of hard corals ranked a medium-priority bioindicator for use in long- and short-term monitoring programmes (Table 3). In contrast, increases in macroalgal cover associated with declines in water quality are well known, with perhaps the best evidence being provided by the shift in the coral reef assemblage from one dominated by hard corals to a high cover of macroalgae due to the introduction of sewage into Kaneohe Bay (Smith et al. 1981). Changes in macroalgae have a high specificity, and are highly relevant in the public perception, making it a high priority for use in long-term monitoring programmes provided the confounding influences of seasons and grazing pressure of herbivores are taken into consideration (Hughes et al. 2007).

Community structure

Community structure is defined as the relative abundance of different taxa at a location. Links are often drawn between coral community structure and exposure to water quality because species’ abundances generally reflect environmental conditions (e.g. van Woesik et al. 1999; Table 2). For example, the community structure of macroalgae may be explained by responses to water quality. Fabricius et al. (2005) reported strong gradient effects in macroalgae at different spatial scales, with greater relative abundances of Rhodophyta and Chlorophyta on coastal reefs adjacent to a region exposed to elevated loads of terrestrial runoff, compared with a region with lower nutrient and sediment inputs. This measure, however, has a variable response time, with rapid changes after acute disturbances that result in mortality, and gradual shifts in response to chronic water-quality changes (Brown et al. 1990). Being a multivariate measure, the interpretation of changes in community structure requires knowledge of the biology of the individual species involved, as well as the disturbance history of a location. Thus, community structure ranked a medium-priority bioindicator for use in long- and short-term monitoring programmes (Table 3). The methods to monitor coral community structure are established, and are based on standard monitoring techniques such as photo-transects (English et al. 1997).

Taxonomic richness

Taxonomic richness is defined as the number of taxa within a community. The taxonomic richness of hard corals, octocorals and macroalgae gradually change along water quality gradients in several coastal regions of the Great Barrier Reef (van Woesik et al. 1999; Fabricius et al. 2005). For example, the taxonomic richness of octocorals with zooxanthellae declined with increasing turbidity and sedimentation (Fabricius et al. 2005; Table 2). Existing studies suggest that the taxonomic richness of hard corals, octocorals and macroalgae have a high specificity to changes in water quality. In addition, decreases in taxonomic richness will occur in response to both chronic and acute reductions in water quality, either because of rapid mortality or reduced settlement success of the most susceptible species. Taxonomic richness is, therefore, considered a high-priority bioindicator for use in long- and short-term monitoring programmes (Table 3), although extensive and ongoing observer training is required to identify and quantify the different taxa and cross calibration is necessary to achieve any consistency. As with community structure, the methods to monitor taxonomic richness are established (English et al. 1997), but require high-level taxonomic expertise.

Maximum depth of coral-reef development

The maximum depth of seagrass distribution has been used as a bioindicator of estuarine health (Abal and Dennison 1996), and a similar approach has been used for coral communities to infer changes in water quality on coral reefs (Kleypas 1996; van Woesik et al. 1999; Cooper et al. 2007). Coral growth and distribution is determined by available irradiance (Yentsch et al. 2002), and the maximum depth of coral reef development can be defined as the zone of transition from zooxanthellate hard corals to azooxanthellate filter feeders along a depth gradient. On the Great Barrier Reef, the maximum depth of coral-reef development increased almost fivefold within a group of inshore islands along a gradient from low irradiance and elevated nutrients at nearshore locations, to high irradiance and low nutrients at outer locations (Cooper et al. 2007; Table 2). The response time of changes in maximum depth of coral growth is most likely to be on a timescale of months to years. Thus, the maximum depth of coral-reef development ranked a high priority for use in long-term monitoring, but was a low priority for use in short-term monitoring programmes (Table 3). The maximum depth of coral reef development can be quantified by visual estimates or with transects along a depth gradient.


Each of the bioindicators identified here has a different response time (i.e. from near instantaneous to years) and specificity to changes in water quality, and the final choice of bioindicators must depend on the specific objective and timeframe available for a monitoring programme. Responses that are specific to disturbances such as nutrient availability, sedimentation and turbidity and that demonstrate a rapid response time, can provide an early warning of changes in water quality. In particular, sublethal bioindicators that reflect rapid changes at the genetic/colony level of selected species may be among the most effective ways to pre-empt mortality of corals and impacts to the wider community. In addition, understanding the processes acting within colonies and among the life-history stages of a selected species will provide a better understanding of the consequences of changes in water quality on populations and coral-reef communities. The comparison of responses from a composite of bioindicators will provide the most useful information on the status and trends of reef ecosystems. As the extent of water-quality degradation increases, so does the scale at which the responses are manifested, and the time taken for the system to return to its previous state when the stressors are removed. Responses to small changes in water quality may be best measured by focusing on rapid sublethal effects at the genetic/colony level in the most susceptible species or life stages; when the stressors are removed, then the condition of these organisms may rapidly return to their previous state. With further decreases in water quality, responses will be measured as partial mortality or mortality of organisms in an increasing number of species or life stages, which becomes evident at the scale of populations and communities, and these signals will remain measurable for months to years after the stressors are removed. Finally, extreme decreases in water quality produce widespread mortality across a range of organisms that will be evident at the level of entire communities and as a shift in ecosystem state; the time taken for the response to be manifest will vary between chronic or acute events, but recovery, when the stressors are removed, is likely to occur slowly. This is best conceptualised when the bioindicators are represented against increasing levels of stressors, from sublethal stress to mortality (Fig. 2). Exposure to a low-level stress will first invoke a response at the genetic and colony level, such as symbiont photophysiology and coral brightness. As stress increases, either in terms of duration or intensity, responses at the population and community level may become evident through reduced juvenile densities, changes in the community structure, through the loss of susceptible species, increased macroalgal abundances, reduced abundances of large symbiont-bearing foraminifera, and a reduction in the maximum depth of coral-reef development. Response time is, therefore, a critical criterion that underpins bioindicator selection in any environmental monitoring. Moreover, exposure to changes in water quality may invoke different responses in some groups, e.g. elevated nutrient concentrations and turbidity decreases coral brightness, while sedimentation stress increases brightness (through bleaching) on upward facing surfaces. Hence, important information on the specific forms of stress can be gained by investigating the direction and magnitude of responses relative to reference conditions.

Fig. 2
figure 2

Conceptual model of coral bioindicators to indicate increasing exposure to the key components of water quality. Responses are presented in increasing order of effect from stress to mortality resulting from increasing levels of stressors. Responses will depend on both the magnitude and duration of changes in the levels of stressors (e.g. Kuntz et al. 2005). All the responses will first be evident at the genetic/colony level and then in the wider community. Sublethal responses, therefore, may pre-empt more severe effects at the population and community level and can be used to describe shifts in ecosystem condition from healthy (green) to degraded (red) conditions

This review has identified 11 measures as high-priority bioindicators that should be considered for inclusion into long-term monitoring programmes. These are: symbiont photophysiology, colony brightness of corals, tissue thickness and surface rugosity of massive corals, skeletal elemental and isotopic composition, abundance of macro-bioeroders, coral recruitment, micro- and meiobenthic bioindicators such as foraminifera, macroalgal cover, taxonomic richness of corals and the depth of coral-reef development. For short-term monitoring programmes such as required for EIA that aim to quantify effects of acute disturbances on coral communities, eight priority bioindicators were identified as: symbiont photophysiology, colony brightness of corals, tissue thickness of massive corals, skeletal elemental and isotopic composition, partial mortality, micro- and meiobenthic bioindicators, coral recruitment and taxonomic richness of corals. The difference between the two suites of bioindicators primarily reflects differences in the response time. Short-term monitoring programmes would be best served with sublethal bioindicators or those evident in the most susceptible species, which have rapid response times. In the case of acute changes in water quality, the exceedence of some pre-determined threshold for a bioindicator may trigger a management response that could be used to limit further impacts. In contrast, monitoring the biological consequences of chronic exposure to lowered water quality as part of long-term monitoring programmes would be best based on bioindicators with slower response times coupled with a suite of sublethal bioindicators to provide an early indication of ecological change. Pomeroy et al. (2004) identified a range of measures that also included taxonomic richness and coral recruitment as priority biophysical indicators for assessing the effectiveness of marine protected areas. However, this review has also identified that sublethal measures have important applications in monitoring programmes. For either short- or longer-term monitoring programmes, assessments should be based on deviation from reference values (historic data or reference locations), while also controlling for other changes in physical and biological conditions that affect the specificity and variability of the response. In both cases, an assessment framework, such as the one suggested here, should provide the basis for the selection of a suite of bioindicators to be used when assessing the consequences of changes in water quality around coral reefs.