Uncovering the volatile nature of tropical coastal marine ecosystems in a changing world

Introduction

Biogenic volatile organic compounds (BVOCs) are a group of chemicals with high volatility under ambient temperatures that are produced by organisms for key ecophysiological processes including excess energy dissipation and thermotolerance, and as antioxidants. Although BVOCs encompass a wide range of compounds with broadly differing characteristics and impacts, two climatically important BVOCs in particular have received global attention: (i) dimethyl sulphide [DMS; (CH₃)₂S], which is almost exclusively produced by marine organisms, and is the most abundant and best studied BVOC in the marine environment (Wingenter et al., 2004; Nightingale & Liss, 2006; Carpenter et al., 2012) and (ii) isoprene (2-methyl-1,3-butadiene; C₅H₈), which is the most abundant BVOC in the terrestrial environment (Sharkey, 2013), although there is increasing evidence of a significant role in the world's oceans (Meskhidze & Nenes, 2006; Acuña Alvarez et al., 2009; Arnold et al., 2009; Shaw et al., 2010; Exton et al., 2012, 2013). Whilst many other trace gases exist in marine systems (e.g. other non-methane hydrocarbons, halocarbons and nitrogen-containing compounds; Carpenter et al., 2012) their abundance is typically low compared to DMS (marine) and isoprene (terrestrial) and they are less well-studied. Consequently, herein we focus on DMS and isoprene and the role they play in marine ecosystems.

The main precursor to DMS is dimethylsulphoniopropionate (DMSP), a secondary metabolite that has multiple physiological functions in marine algae (Stefels et al., 2007). DMSP can be degraded by several prokaryotic and eukaryotic catabolic pathways, some of which result in DMS (Steinke et al., 2007; Todd et al., 2007; Curson et al., 2011). Here, we include DMSP in our discussion because of the important role it plays as a precursor to DMS formation although it is not considered a BVOC in its own right. DMS is the main source of global biogenic sulphur emissions (Charlson et al., 1987), making it a key contributor to the global sulphur cycle. In the oceans, DMS is traditionally considered to be the major BVOC impacting on climate, particularly playing a vital role in cloud condensation nuclei (CCN) formation, but importantly increasing CCN density and thus increasing cloud albedo (Charlson et al., 1987; Vallina & Simó, 2007). However, recent work has suggested that additional marine-sourced BVOCs also have major impacts on atmospheric processes over both coastal and open-ocean regions (Gantt et al., 2010). Notably, in areas of high primary productivity, isoprene can significantly affect ocean climate (Gantt et al., 2009) by modulating shallow marine clouds through the production of secondary organic aerosols (SOA) (Meskhidze & Nenes, 2006). However, the influence of BVOCs is far from restricted to atmospheric chemistry as they serve a number of roles at different ecological scales ranging from modifying the stress tolerance of specific organisms (Sunda et al., 2002; Vickers et al., 2009a,b), to influencing community level trophic interactions as infochemicals (Gershenzon, 2008; Breckels et al., 2011; Nevitt, 2011). Thus, BVOCs influence marine systems over local, regional and global scales.

Current understanding of marine BVOC production and the ecological roles that BVOCs play is extensive in the case of DMS, and emerging in the case of isoprene. However, some systems remain severely understudied, including the most biodiverse of marine systems, corals reefs, and their associated mangrove and seagrass ecosystems. Such a lack of BVOC knowledge is surprising given the global attention that these systems attract, together with their well-documented biological and socioeconomic importance (Hughes et al., 2003). In fact, research into isoprene and DMS since the 1980s have resulted in 924 publications for oceanic ecosystems and phytoplankton, compared to only 31 from tropical coastal marine ecosystems (as of 22nd May 2014, ISI Web Of Science, Thompson Scientific). However, several lines of evidence suggest that tropical coastal marine ecosystems should be ‘hot spots’ for BVOC emissions and the roles they play: (i) highly productive per unit area (Hatcher, 1988); (ii) supported by primary producers that are among the highest BVOC producers on the planet (e.g. Broadbent et al., 2002); (iii) dominated by shallow water autotrophs that routinely experience physiological stress likely to drive BVOC production (Baker et al., 2008; Suggett & Smith, 2011) and (iv) exhibit complex ecological interactions that are known to be strongly mediated by biologically derived chemical products (e.g. DeBose et al., 2008; Borell et al., 2013).

Coral reefs and their connected habitats are already undergoing dramatic shifts in biodiversity and productivity as a result of environmental change and anthropogenic pressures (Hoegh-Guldberg et al., 2007). Whilst the functional role of BVOCs within these complex ecological systems are not yet fully understood, our existing knowledge of other marine and terrestrial environments would suggest that they serve a number of fundamental, but as yet unexplored, functions at different ecological and climatic scales; indeed expanded research on the operation of BVOC-based interactions within reef systems, as per other ecosystems, would likely reveal a new ‘infochemical network’ that could explain a number of unresolved issues in ecological functioning. Here, we combine our current understanding of BVOCs with our ecological knowledge of tropical coastal ecosystem function to provide a perspective on potential roles for BVOCs, and thus provide a framework for future research activities. In doing so, we highlight the critical need to address the existing lack of knowledge concerning current BVOC-driven interactions and processes that would support the considerable research efforts focused on physiological, ecological and biogeochemical and climatic studies, particularly those addressing system responses to rapid environmental and climatic change.

Tropical marine BVOC emissions and a changing planet

Marine algal BVOC production is known to be dominated by DMS (Simó, 2001; Broadbent et al., 2002; Nightingale & Liss, 2006), whilst vascular plants tied to a terrestrial existence generally produce isoprene in greatest abundance (Sharkey & Yeh, 2001) (Table 1). Consistent with this notion, corals produce DMS (e.g. Broadbent et al., 2002; Broadbent & Jones, 2004), largely via their symbiotic algae (e.g. Van Alstyne et al., 2009; Steinke et al., 2011) but also from the cnidarian host itself (Raina et al., 2013). In contrast, mangroves have been shown to produce significant amounts of isoprene (Bandaranayake, 2002; Barr et al., 2003); however, it is now known that corals' symbiotic algae of the genus Symbiodinium are also significant isoprene producers (Exton et al., 2013). Similarly, microphytobenthic (MPB) communities are abundant isoprene producers in temperate systems (Exton et al., 2012), and thus presumably tropical MPB should also contribute to isoprene emissions. Other BVOC-emitting organisms in tropical coastal systems probably do so indirectly; for example sponges are believed to emit DMS through their ingestion of DMSP-containing food sources rather than through active sponge-driven biosynthesis (Van Alstyne et al., 2006). Seagrasses have been identified as a source of DMSP, although it appears that this is largely due to their epiphytes as would be expected in DMSP-limited vascular plants (Dacey et al., 1994).

Of course, these general findings are based on measurements from just a few representative species within each system, but such group-specific trends in BVOC production suggest that the tropical coastal environment experiences a unique cocktail of BVOCs compared to exclusively pelagic or terrestrial systems. It follows that tropical coastal environments likely exhibit a BVOC production gradient stemming from DMS-dominated subtidal benthic communities such as corals and macroalgae through to littoral and supralittoral isoprene-dominated communities of vascular plants such as mangroves and seagrasses (Fig. 1). Phytoplankton and MPB species, which produce both DMS (e.g. Simó, 2001) and isoprene (e.g. Exton et al., 2013), would persist throughout this gradient, although their abundance and productivity tend to be higher in the nutrified waters of mangroves rather than oligotrophic waters of coral reefs.

Most studies of BVOC production in tropical coastal ecosystems to date have focused on reef-building corals. Concentrations of DMS have been shown to increase in both the water and air close to coral reefs (Andreae et al., 1983; Jones & Trevena, 2005; Jones et al., 2007; Swan et al., 2012). Of all coral species tested, highest DMSP production has been observed from species of Acropora (up to 7.66 mmol m⁻² reef) (Jones et al., 1994; Broadbent et al., 2002; Tapiolas et al., 2013), and concentrations of both DMS and DMSP appear particularly high in coral-produced mucus ropes (up to 19 and 54 μM respectively) (Broadbent & Jones, 2004). Whilst these earlier studies provide compelling evidence for corals being a major source of DMSP and DMS, there remains some debate as to the specific source of DMSP within the coral holobiont. Many studies have presented results suggesting that corals' symbiotic microalgae, Symbiodinium spp., are the sole source of DMSP in cnidarian-algal symbioses (Van Alstyne et al., 2009), and that Symbiodinium density and DMSP production are significantly positively correlated (Van Alstyne et al., 2006). However, more recent evidence from asymbiotic juvenile corals demonstrates an important and possibly dominant role of the cnidarian host animal (Raina et al., 2013). Postsettlement increases in DMSP production of 54% were followed by further increases of 76% when exposed to thermal stress. These authors discovered several orthologues of algal genes responsible for the biosynthetic pathway of DMSP, suggesting a clear potential for the coral animal to produce DMSP independent of its algal symbionts, although no clear mechanism for host production was given.

Corals clearly have a major role in DMS production; however, less is known as to whether they produce isoprene (but see Exton et al., 2013 for Symbiodinium spp.). Similarly our knowledge of corals vastly outweighs that of other key photosynthetic benthic organisms. However, previously unpublished field data from coral reefs in Indonesia presented here, in which a range of organisms were analysed using chemiluminescence-based sensing, suggests that BVOC production is higher in corals than other primary producers such as macroalgae and seagrasses (see Fig. 2 and legend for full methodological approach). Whilst this chemiluminescence technique does not discriminate between DMS and isoprene when left unfiltered (Exton et al., 2010; Green et al., 2012), it does provide a broad proxy for total BVOC production and highlights novel trends that carry clear ecological implications. It also presents the first assessment of BVOC production rates across multiple taxa within the same ecosystem using a standardized technique, and therefore offers a valuable insight into the relative importance of different tropical marine primary producers in terms of BVOC production.

Firstly, BVOC production was strongly associated with growth form and species. As expected, based on existing research into DMSP in corals (Jones et al., 1994; Broadbent et al., 2002; Broadbent & Jones, 2004), the highest production was from a species of Acropora (A. nobilis), a complex branching species common to the Indo-Pacific region. The lowest rates were found in three submassive coral species (Porites nigrescens, Stylophora pistillata and Pocillopora eydouxi, representing <7% of the A. nobilis rates based on Chl a normalized data), and intermediate production rates for three massive species (Favia speciosa, Diploastrea heliopora and Porites lutea: <24% of the A. nobilis rate) and one foliose species (Montipora foliosa: <40% of the A. nobilis rate). In fact, the intermediate rates for the massive species were still relatively high compared to many other corals and macroalgae tested here. Although production data for corals were highly variable, key patterns of variation between species exceeded this within-species variation sufficiently for a broad discussion of taxonomic trends. Consequently, severe environmental impacts that frequently favour selection of massive over branching species, such as thermal bleaching (e.g. Suggett & Smith, 2011), ocean acidification (e.g. Fabricius et al., 2011) and storm action (e.g. Fabricius et al., 2008), and lead to ‘reef flattening’, would be expected to maintain relatively high BVOC production per surface area of coral. However, the loss of fast-growing, complex branching species, such as Acropora spp., that contribute high coral tissue surface area per unit area of reef substrate would likely bring about an overall system-wide decrease in gross BVOC production, in agreement with Broadbent et al. (2002).

Secondly, BVOC production by macroalgae was at least 45% less than the A. nobilis production when normalized to Chl a, and 12% less when normalized to surface area. In most cases, macroalgae were also lower producers than massive and foliose corals. Therefore, conditions that favour enhanced spatial coverage of macroalgae, such as ocean acidification (Fabricius et al., 2011; Johnson et al., 2012), hypernutrification (McManus et al., 2000) and a loss of grazers (Mumby et al., 2007), could decrease overall BVOC production per unit reef area. The well-documented phase shifts away from reef-building coral-dominated systems would thus likely result in a net loss of BVOC production, where the highest producers are systematically lost from reefs, although this reasoning takes no account of adaptation to stress events (see Jones et al., 2014).

How environmental change might affect BVOC production within mangrove systems is perhaps less clear. Whilst more CO₂-enriched and warmer atmospheric conditions are expected to increase productivity (Ball et al., 1997), it seems likely that isoprene production will decrease due to a self-regulating inhibition of isoprene synthesis common to vascular plants and driven by the increase in atmospheric CO₂ (Arneth et al., 2007). Mangrove loss via deforestation and coastal development is accelerating at an alarming rate (Duke et al., 2007), and when this loss is combined with the inhibition driven by increased atmospheric CO₂, it may be that any temperature-driven rise in isoprene emission is negated. That said, although coastal nutrification caused by increased input (both direct to marine systems or indirectly through riverine input) will reduce coral productivity and health (Vega Thurber et al., 2014), the contribution of phytoplankton to BVOC emissions would inevitably increase in line with nutrient stimulated increases in biomass and productivity (e.g. see Fabricius et al., 2013). Consequently, any loss of BVOC production from corals and mangroves could partly be offset by increased microalgal production. Clearly much research effort is needed to fully quantify the contribution of different taxonomic and functional groups to ecosystem BVOC emissions, and without sufficient data it is impossible to confidently state how future ecological and climatic change will impact on BVOC emissions. Furthermore, along with increased knowledge of baseline production rates amongst organisms and species, there is also a critical need to develop studies that consider how perturbed environmental conditions regulate BVOC production from organism to ecological scales.

Expanding this issue to climatic scales raises the interesting question of feedback scenarios, specifically whether the tropical marine biosphere possesses the capacity to either regulate or exacerbate the effects of climate change via their BVOC emissions; an important question when considering the resilience of coral reef ecosystems. The well-publicized CLAW hypothesis (Charlson et al., 1987) describes a classic negative feedback scenario, whereby increases in atmospheric BVOC emissions drive increased shading through cloud formation and increased albedo (Vallina & Simó, 2007). This has been supported by a proposed ‘ocean thermostat’ in the Western Pacific Warm Pool (WPWP) region, based on this cloud-SST mechanism maintaining water temperatures within optimal ranges for coral reefs (Kleypas et al., 2008). A similar potential negative feedback mechanism has been proposed for the Great Barrier Reef based on the high DMS emission rates over coral reefs (Fischer & Jones, 2012). However, Fischer & Jones (2012) also observed an almost complete shut-down of atmospheric DMS emissions from laboratory corals under temperature and light stress conditions, suggesting a positive feedback via reduced atmospheric DMS fluxes is also a possibility.

Role of BVOCs in regulating stress susceptibility

A key consideration for research into community-level BVOC production is the distinction between gross production and net emission. Gross BVOC production relates to biosynthesis alone whilst net BVOC emission depends on internal processes at the organismal level, and the contribution of other associated organisms such as epibionts and bacteria at larger ecologically relevant scales. Thus, examining production properties in organisms ex situ and isolated from their community may not only be misleading in terms of actual BVOC emission rates, but may not represent the key dynamics of production mediated by ecological interactions. However, such studies do give important baseline values from which future changes can be assessed.

Important observations from corals have recently demonstrated that microbial communities associated with coral surface tissue and mucus metabolize DMSP and other complex coral metabolites (Raina et al., 2009, 2010; Bourne et al., 2013). Other studies describe isoprene-degrading microbial communities within tropical coastal seawater (Acuña Alvarez et al., 2009). These communities that facilitate degradation processes will ultimately determine the proportion of benthic-sourced BVOCs found in the water column and, subsequently, the atmosphere, and therefore should be considered important components of the tropical coastal BVOC cycling pathway (Fig. 1). Physical processes, notably tidal emersion (exposure during low tides), that act to reduce the contribution of pelagic consumption and increase the contribution of benthic production may act to enhance BVOC emissions directly to the atmosphere. Furthermore, tidal emersion will often simultaneously subject benthic producers to environmental stress (e.g. elevated light and temperature), thereby stimulating greater gross production of BVOCs during low tides in the day when photosynthesis is greatest on reef flats. Such BVOC contributions during emersion will not be coupled with pelagic processes and hence largely avoid the effects of degradation, leading to greater net emissions. This point has been illustrated by increased atmospheric DMS concentrations during daylight low tides on reef flats in the Great Barrier Reef (Jones et al., 2007). Similar findings have also been reported for isoprene concentrations in water above MPB-dominated mudflats in a temperate estuary (Exton et al., 2012).

Cellular and organism-scale physiological benefits of both DMSP and isoprene production are well-documented, and their role within the synthesizing organism is of significant ecological importance. DMSP is a well-described antioxidant in marine algae (Sunda et al., 2002), while isoprene research in terrestrial systems has identified a clear role in both thermotolerance (Sharkey & Singsaas, 1995) and as an antioxidant (Loreto & Velikova, 2001; Vickers et al., 2009a,b). Specific perturbations in the environment result in dramatic changes in BVOC production that reflect a role in either stress susceptibility or stress tolerance:

At the broadest level, seasonal changes in environmental conditions (most notably temperature and light) acting on natural coastal marine communities have been shown to positively impact production of both DMS (Jones et al., 2007) and isoprene (Exton et al., 2012). Such responses likely mirror those operating at the organism scale. For example, a temperature- and light-dependent response of isoprene production rates in phytoplankton cultures has previously been described (Shaw et al., 2003), supporting similar work in higher plants (Sharkey & Loreto, 1993). Similarly, DMS water concentrations over a coral reef have been shown to increase with temperature up to 30 °C, after which concentrations decreased (Jones et al., 2007). This latter finding was supported by chamber experiments using Acropora formosa that demonstrated a decline in DMSP production above the bleaching threshold of 31 °C (Jones et al., 2007), and using Acropora intermedia that showed a decrease in atmospheric DMS production with temperature increases to 2 °C above ambient (Fischer & Jones, 2012). Another study demonstrated a five-fold increase in DMSP concentrations in corals under thermal stress (Garren et al., 2014), while further research has identified in situ increases in DMSP production in Acropora intermedia when water temperatures increased by only 2 °C (Jones et al., 2014). Interestingly, elevated DMSP production even 6 months after a bleaching event was observed, suggesting an adaptive upregulation in BVOC synthesis to protect against future stress (Jones et al., 2014). In perhaps the most comprehensive study to date involving corals, production of dimethylated sulphur compounds (DMS, DMSP and dimethyl sulphoxide; DMSO) were shown to increase under antioxidant stress caused by variations in temperature, light, salinity and air exposure (Deschaseaux et al., 2014), with each sulphur species exhibiting different responses to each environmental cue. Similar responses have also been shown for other groups of reef organisms, for example an increase in DMSP has been shown in response to diel variation in water temperature and light in Red Sea coralline algae (Burdett et al., 2014).

In the case of ocean acidification, evidence is currently divided, with DMS(P) reported to increase under elevated CO₂ levels in natural temperate pelagic phytoplankton communities (Hopkins & Archer, 2014) and in periods of low carbonate saturation state (Burdett et al., 2013); such observations may potentially reflect the link between photosynthesis-derived methionine production and DMSP synthesis (Stefels, 2000; Burdett et al., 2014), where photosynthetic rates can be enhanced by elevated pCO₂ (Rost et al., 2008; Suggett et al., 2013). However, phytoplankton DMS(P) has also been shown to decrease with reduced pH in mesocosm experiments (Hopkins et al., 2010; Six et al., 2013). For isoprene, the impacts of ocean acidification on production rates have yet to be addressed. Research has indicated a suppression of isoprene synthesis by vascular plants under increased atmospheric CO₂ (Arneth et al., 2007; Possell & Hewitt, 2010), although how this translates to altered emissions from mangrove plants is unknown. Reef-forming corals and seagrasses have been the focus of much ocean acidification research to date, but this has not been linked to their role as BVOC producers, and at present we are limited to making inferences based on community-scale alterations where interspecific variations in BVOC production capacity are known to exist (see above).

Despite the increasingly intensive research on how environmental and climate stressors will influence BVOC production, the exact connection between (and underlying mechanisms regulating) physiological health and BVOC production are still unclear. Even so, some published data on BVOC production by Symbiodinium suggest a unique hypothesis in terms of the role of BVOCs in regulating stress susceptibility:

Thermally induced coral bleaching is perhaps the most significant threat to the future of these vital ecosystems (Hoegh-Guldberg et al., 2007; Suggett & Smith, 2011). The exact physiological mechanism(s) that leads to high temperature- (and light-) driven coral bleaching is still under debate, but it is generally acknowledged to involve an accumulation of reactive oxygen species (ROS) produced by Symbiodinium and/or the coral host (e.g. Baker et al., 2008; Weis, 2008). One intriguing proposal is a decrease in thylakoid membrane stability as a result of increased H₂O₂ release (Tchernov et al., 2004; Smith et al., 2005; Suggett et al., 2008). Recent investigations into BVOC production by genetic variants of Symbiodinium spp. have suggested significantly higher ambient production of isoprene in thermally sensitive variants (Exton et al., 2013), compared to relatively consistent concentrations of DMSP (Steinke et al., 2011) (Fig. 3). At first glance, this potentially disproves both the DMSP-oxidative stress theory (above), and a potential role of isoprene in stress. Whilst these measurements took place under optimal growth conditions it is reasonable to expect their production to be (i) higher under anomalous environmental conditions as shown with isoprene in higher plants (e.g. Sharkey & Loreto, 1993) and (ii) taxonomically variable as seen in oceanic phytoplankton (Shaw et al., 2010; Exton et al., 2013). Initial evidence points towards significant upregulation of BVOC production in both algae and corals under stress conditions (Shaw et al., 2003; Jones et al., 2007, 2014; Garren et al., 2014; Deschaseaux et al., 2014), but what remains to be seen is whether organisms with high production rates under ambient conditions have the same upregulation potential as those with lower background rates, thus raising the key question of background production vs. upregulation capacity.

If those coral-symbiont relationships with enhanced thermal tolerance possess enhanced isoprene upregulation capacity, this links well with the known function of isoprene in providing thermotolerance to higher plants. Isoprene molecules have been shown by modelling to partition into the centre of the cellular lipid bilayer, bringing about an increase in the level of ordering of lipid tails (Siwko et al., 2007), thereby stabilizing the membrane against the adverse effects of high temperature episodes. Despite lower isoprene production rates in algae compared to many vascular plants, only small quantities could provide a substantial benefit to organisms under elevated temperatures, even if production rates are not high enough to also afford significant antioxidant benefits. Instead, the more highly concentrated DMSP would provide the additional scavenging required to negate the effect of ROS released under elevated temperatures. Interestingly, patterns in isoprene production amongst Symbiodinium variants with different stress tolerances (Fig. 3) appear to be generally consistent with the coral BVOC data reported in Fig. 2; specifically higher BVOC production in branching corals and lower production in massive/submassive corals, which can be broadly considered less or more heat-stress tolerant, respectively (e.g. Suggett & Smith, 2011; van Woesik et al., 2011). Such potential connections between BVOC production and stress susceptibility are indeed intriguing but clearly require much more intensive investigation and particularly a focus on the dynamics of BVOC regulation, and possible interplay between different BVOC species, in response to changing environments.

Infochemistry and a role in community interactions

Much of the attention on BVOCs has focused on their roles in climate and biogeochemistry, and to a lesser extent ecophysiological function to stress response; however, an increasing body of literature is highlighting their ecological importance as signalling molecules. On land, isoprene is well-established as an infochemical (e.g. Gershenzon, 2008; Laothawornkitkul et al., 2008), whereas in the oceans DMS and DMSP are known to be similarly important (e.g. DeBose et al., 2008; Garren et al., 2014). Signalling roles can take several forms, ranging from defence against biotic stress to catalysts for interspecific interactions, and has begun to greatly improve our understanding of the mechanisms behind complex ecological processes (e.g. DeBose et al., 2008; Gershenzon, 2008; Laothawornkitkul et al., 2008; Garren et al., 2014). Once again, the vast majority of studies into BVOCs as infochemicals are weighted towards terrestrial and non-tropical marine environments. However, the high diffusion potential of BVOCs in aquatic environments makes them ideal signalling molecules. With the shear scale of biodiversity found in tropical coastal systems, and the complexity of ecological interactions taking place, it is highly plausible that a range of infochemically driven ecological processes occurs.

DMSP and DMS are well-described chemo-attractants to a diverse group of organisms across the taxonomic spectrum. At the most basic level, DMSP has been shown to attract phytoplankton, heterotrophic bacteria, as well as both bacterivorous and herbivorous microzooplankton (Seymour et al., 2010). However, zooplankton grazing increases DMS emissions (Wolfe & Steinke, 1996), which also acts as a foraging cue for higher order predators including African penguins (Cunningham et al., 2008), seabirds (Nevitt et al., 1995), loggerhead turtles (Endres & Lohmann, 2012), and seals (Kowalewsky et al., 2006). These species use DMS to locate areas of high marine productivity where prey abundance is likely to be highest, in what are termed multi-trophic interactions. Some of these larger predators are important transient visitors to coral reefs, forming a significant component of their associated diversity. The extremely high production of DMSP and DMS on coral reefs (Broadbent et al., 2002; Broadbent & Jones, 2004) must thus be an extremely attractive proposition for those species possessing detection capabilities.

While charismatic megafauna perhaps understandably have a more prominent presence in the headlines, fish are also known to demonstrate positive responses to DMSP, helping to explain some aggregation and potentially migratory behaviours. Planktivorous reef fishes will aggregate in response to increased DMSP concentrations in the water column at naturally relevant concentrations (DeBose et al., 2008), although this is not the case for all species, suggesting taxonomic variability in either detection ability or behavioural response. Transient reef visitors also respond to elevated DMSP and DMS on the reef, including pelagic fishes (specifically Carangidae) and squid (DeBose & Nevitt, 2008). Further experimental studies have demonstrated a positive effect of DMSP on swimming behaviour in juvenile Carangids (DeBose et al., 2010). Thus, a network of signalling likely operates across trophic levels, which is highly sensitive to alterations in environment as well as in community structure; factors such as broad overfishing and the targeted removal of species have the potential to disrupt important multitrophic interactions, with currently unknown effects on BVOC release.

Although it has been suggested that isoprene may also act as a chemoattractant (Schnitzler et al., 2010), there are several examples of a more varied role. Naturally emitted isoprene deters caterpillars from grazing (Michelozzi et al., 1997; Laothawornkitkul et al., 2008), as well as deterring bacterivores in soil (Fall & Copley, 2000), while it also repels parasitic wasps acting as plant bodyguards which tri-trophically promotes herbivory (Loivamaki et al., 2008). This latter role has even led to the suggestion that some plants may use increased isoprene production to ‘wage war’ on neighbours that are more susceptible to herbivory than themselves (Gershenzon, 2008). With the comparably low concentration of isoprene in marine systems, its importance as an infochemical may at first glance be less significant than for DMS and DMSP. However, with the low diffusion potential of BVOCs in water compared to air, strong concentration gradients will emanate from source organisms. Thus, the previously discussed potential for an isoprene-DMS gradient to exist across the tropical coastal habitat continuum raises important questions as to the potential significance of total BVOC suite, spatial variability and effects on regional climate variability (Fischer & Jones, 2012). Numerous localized migration patterns exist amongst reef fish species, including several involving movements between intertidal mangrove and seagrass environments and the subtidal reef proper (Nagelkerken et al., 2000; Mumby, 2006). However, the navigational mechanism behind these migrations is yet to be understood, and the ability to sense a DMS(P)-isoprene gradient may hold the key to fish movement behaviour between a DMS and DMSP-dominated BVOC suite at the pelagic-interface to an isoprene-dominated BVOC suite at the terrestrial-interface.

Despite their clear importance in driving complex food webs, the concept of BVOCs as infochemicals extends far beyond mere prey location. Some reef macroalgae are known to use a suite of secondary metabolites as chemical defence mechanisms against herbivory (Hay et al., 1987). Similarly, some sponges (likely via their microbiome) synthesize chemicals that are unpalatable to fishes in order to deter predation (Kubanek et al., 2002; Pawlik et al., 2002). This suggests that biogenic compounds play important functions in determining focused feeding patterns, while the fact that this mechanism does not appear to be ubiquitous across species suggests a role in interspecific competition in complex coral reef communities. A more recent study demonstrated a negative relationship between feeding preference in herbivorous reef fishes and the density of macroalgal stands (Hoey & Bellwood, 2011), a pattern also found in New Zealand kelp forests (Taylor & Shiel, 2010). Several ecological reasons were proposed to explain this finding and, although the potential involvement of BVOC signalling was not addressed, findings from other ecosystems suggest it could provide a suitable explanatory mechanism. Regardless, it raises worrying questions about the future prevalence of phase shifts on coral reefs if one of the more likely routes to recovery (i.e. herbivory) is suppressed as total macroalgal cover increases. With increases in water temperature likely to drive elevated BVOC production from reef phototrophs, as discussed previously, this effect will be magnified, leading to increased suppression of herbivory. Another recent study adds to this concern, identifying chemically induced reduced fitness in corals caused by terpenes produced by reef macroalgae (Rasher et al., 2011), further aggravating the continued deterioration of many coral reefs.

Beyond the process of feeding, DMS and DMSP concentrations in the water column are known to increase during coral spawning episodes (DeBose & Nevitt, 2008), and chemical cues are known to play a part in reef fish larval settlement (Sweatman, 1988). This latter role raises the possibility of a decrease in successful larval settlement on degraded reefs through the loss of high BVOC producing but environmentally sensitive taxa such as those of Acropora, which would have implications for the life cycles of reef-associated animals as well as the provision of important economic and food security benefits to local human communities if fishery recruitment patterns were to diminish. However, perhaps one of the most fascinating infochemical roles of BVOCs so far identified is in coral disease transmission. Laboratory experiments have demonstrated that the coral pathogen Vibrio coralliilyticus uses high DMSP concentrations in coral mucus to locate host colonies (Garren et al., 2014). It is interesting to note that the DMSP was used purely as a navigational tool by the pathogen, as no DMSP catabolism was observed. If other pathogens demonstrate a similar response, and a positive feedback for reef BVOCs occurs, a correlated increase in the prevalence of coral disease is a distinct possibility, at a time when overall health and function will already be compromised by other factors such as increased temperature, ocean acidification and phase shifts in benthic community dominance.

Conclusion

Unresolved complexities associated with determining overall BVOC production, combined with contemporary shifts in community structure and climate predictions for tropical coastal systems, raise important questions as to future climate feedback scenarios. With both DMS and isoprene known to play a role in low level cloud formation (Charlson et al., 1987; Vallina & Simó, 2007; Gantt et al., 2009), and with a correlation between environmental stress and production rates observed in other systems (Jones et al., 2007; Exton et al., 2012), future increases in sea surface temperature could drive a negative feedback loop via increased CCN and subsequent protection against further temperature increases, an idea supported by field data showing a cooling effect of increased cloud cover on coral reef sea surface temperature (SST) (Fischer & Jones, 2012; Leahy et al., 2013). However, the added complication of responses to multiple stressors such as ocean acidification and warming, eutrophication and overharvesting make accurate predictions extremely challenging; indeed, accurately understanding the response of BVOC production will require that experiments begin to address likely interactions amongst such stressors (Arnold et al., 2013; Ban et al., 2014; Rudd, 2014). Even so, the extent to which these feedback mechanisms will operate will ultimately depend on future BVOC production, which in turn is determined by the community composition and specifically either (i) the continued presence of high BVOC producers in significant abundance or (ii) an upregulation of BVOC production amongst those species with currently low production rates in present day environmental regimes. It is clear that our current understanding of BVOCs in tropical coastal marine environments is totally inadequate for making accurate and robust estimates of present, let alone future, predictions of production and their effect on regional climate change. Similarly, while evidence from other ecosystems and the handful of studies that have addressed BVOCs on and around coral reefs (and their neighbouring ecosystems) indicates an important role in both stress response and complex community dynamics via infochemistry, we have barely begun to scratch the surface of the true extent to which DMS, isoprene and many other BVOCs are linked to the health and function of these critical marine ecosystems.