Evidence for water-mediated mechanisms in coral–algal interactions

1. Introduction

Competition for space plays a major role in structuring networks of interacting species [1]. On coral reefs, intense competition occurs between sessile benthic organisms, particularly corals and algae that can be two major components of the benthos [2]. During recent decades, many tropical reefs have transitioned from coral-to-macroalgal dominance [3]. This community shift increases the frequency and diversity of coral–algal interactions and frequently results in coral competitive loss that can drive coral into further decline [4]. Algal overgrowth is facilitated by the combined effects of local (e.g. nutrients, overfishing of herbivorous fish) and global (e.g. climate-induced coral bleaching) stresses [3–7]. These stresses alter existing coral–algal interactions and create new interactions with different algal groups through a concomitant shift in algal composition from low biomass crustose coralline algae (CCA) and short turfs towards dense turfs, fleshy macroalgae, and cyanobacterial mats [4,8–9]. Despite being a well-documented phenomenon, the consequence of such a shift for coral–algal interactions and their underlying mechanisms remain poorly understood.

Interactions between corals and algae typically involve a range of physical, microbial, and chemical mechanisms [2,10]. Direct, physical effects include shading, abrasion, and smothering [11,12]. However, plastic algal mimics, used to simulate physical effects, often produce fewer effects on corals than live organisms, supporting the role of chemistry and metabolic activity [13–15]. Chemicals and microbes can be transmitted by contact- or water-mediated mechanisms. However, the relative importance of contact- versus water-mediated mechanisms on the competition between corals and algae is unclear [10]. Direct contact can facilitate the transfer of hydrophobic allelochemicals and pathogens which are present on algal surfaces [8,15,16]. For example, direct contact with some species of macroalgae, as well as their hydrophobic extracts, has been shown to cause bleaching, tissue necrosis, and/or disease of some corals [13,14,17]. Algae also produce or harbour a number of potentially harmful hydrophilic compounds, including dissolved organic matter (DOM), free-living microbes, and exosomes, which are mobile in the water column [10,18]. These compounds will be elevated within the diffusive boundary layer (DBL) downstream of the algae and spread on to downstream neighbouring corals [10,19,20].

Water-mediated interactions are likely to depend on several physical (e.g. light, flow, topography) and biological (e.g. metabolic activities) processes. In the field, seaweeds placed 3 cm away from corals had no impact on the severity and dynamics of a microbe-generated coral disease [21]. More recently, a two-dimensional diffusion model based on oxygen measurements in the field estimated that the transport of a low molecular weight substance produced continuously at a fixed rate by algae is mainly restricted to the DBL and extends only 1 mm across the border to a neighbouring coral [22]. These studies suggest that hydrodynamics limit the impact of algal exudates in the natural environment and that water-mediated interactions between corals and algae largely operate at the mm-to-µm scale within the DBL. The benthic DBL is a thin film of stagnant, diffusion-limited water surrounding the benthos, whose thickness is determined by flow and microtopography [23,24]. Thinner DBLs promote the exchange of dissolved gases and nutrients for metabolic processes and the removal of wastes [25,26]. Conversely, thicker DBLs may lead to the accumulation of harmful substances and bacteria and localized alterations in oxygen concentrations [22,27]. The chemical microenvironment within the DBL of the coral–algal interface is also likely to depend on the rates of release of chemical and microbial compounds from each competitor, which are species specific. For example, under similar environmental conditions, turf algae produce almost twice as much dissolved organic carbon (DOC) per unit surface area compared with other fleshy macroalgae [18]. This could lead to high concentrations of DOC within the DBL at the coral–turf interface and cause enhanced microbial growth and respiration, particularly that of opportunistic pathogens [28], and subsequently coral mortality [29,30].

Here, we describe field observations and laboratory microsensor measurements used to assess the importance of water-mediated interactions between corals and algae. We characterized the oxygen conditions and DBL thickness at interaction zones between the massive coral Porites and different benthic competitors using oxygen microsensors in a flow chamber and compared our results with competitive outcomes in the field in Moorea, French Polynesia. A flow chamber with recirculating natural seawater was preferred over in situ measurements to minimize the effect of macroscale (cm-to-m) variations in ambient flow, light, and topography. Interactions between massive Porites and the following six benthic competitors were studied: (i) massive Porites (i.e. conspecific), (ii) the CCA species Porolithon onkodes, (iii) thin turf algae (less than 5 mm canopy height), (iv) thick turf algae (5–20 mm canopy height), (v) the red macroalga Amansia rhodantha (Rhodophyta), and (vi) the cyanobacterium Hydrocoleum majus. These competitors belong to functional groups which are known to vary in their competitiveness against corals [10,31] and they can be used in a space for time comparison to examine transitions from coral to fleshy algal dominance.

We hypothesized that the interface between Porites and superior benthic competitors is characterized by thick DBL and high diel oxygen fluctuations, including hypoxic conditions at night. We then tested the effect of flow speed and direction on interactions between Porites and two benthic competitors (thin and thick turf algae), with the assumption that DBL thickness and oxygen fluctuations are reduced with increasing flow speed and when the coral is upstream of the algae. Finally, we estimated the importance of bacterial respiration in generating hypoxic conditions during the night by adding antibiotics, with the hypothesis that microbes cause highly hypoxic conditions at the coral–algal interface.

3. Results

(a) Field study

The dominant benthos on the reef hard substratum was thin turf, followed by corals, macroalgae, and thick turf (figure 1a). Massive Porites was the dominant coral. Colonies interacted mostly with thin turf, followed by thick turf, CCA, and macroalgae (figure 1b). Photographic time series demonstrated that H. majus, A. rhodantha, and thick turf algae rapidly gained space at the interaction zone with Porites (figure 1c). Thin turf overgrew Porites but at a significantly lower rate than thick turf, H. majus, and A. rhodantha. In contrast, Porites gained space when competing against CCA and conspecifics, albeit at very slow rates.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

(a) Composition of the reef benthos (hard substratum). (b) Proportion of massive Porites colony edge interacting with major benthic functional groups. (c) Rates of overgrowth of the benthic competitors over massive Porites spp. The dark grey colour indicates the benthic competitors used in the experiments. Rates of overgrowth were analysed by a one-way ranked based ANOVA (F_5,13 = 34.201, p < 0.001). Letters indicate homogeneous subgroups by Tukey post hoc tests. Data points are mean ± s.e.m.

(b) Experiment 1

The oxygen microenvironment at the interaction interface varied widely between competitors (figure 2; oxygen profiles and complete ANOVA results in electronic supplementary material, figure S2 and table S1). During the day, the interfaces with thin and particularly thick turf were hyperoxic (surface oxygen of approx. 300 and approx. 600 µM, respectively) relative to ambient seawater (approx. 180 µM) and to oxygen concentrations on the coral tissue (approx. 210 µM; figure 2a). Day-time surface oxygen concentrations at interfaces with conspecifics, CCA, A. rhodantha, and H. majus did not differ from the coral tissue. During the night, the interfaces with H. majus, A. rhodantha, and thick turf algae were hypoxic (approx. 50 µM) relative to ambient seawater (approx. 170 µM) and to oxygen concentrations on the coral tissue or at the interfaces with thin turf, CCA, and conspecifics (approx. 110 µM) (figure 2b). Surface oxygen concentrations on apparently healthy Porites tissue (i.e. coral measuring position) during day or night did not vary with benthic competitors.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Day and night surface oxygen concentrations (a,b) and DBL thicknesses (c,d) in interactions between massive Porites spp. and different benthic competitors. Competitors were placed upstream of Porites. Data points are mean ± s.e.m. (n = 5). p-values are from rank-based ANOVAs. See electronic supplementary material, table S1 for complete ANOVA results. Letters indicate significant groupings of benthic competitors (BC) for each of the three measuring positions (MP) by Tukey multiple comparisons tests. Horizontal bars indicate significant groupings of measuring positions for each benthic competitor. Note scale differences on the y-axis for day and night surface oxygen concentrations.

Surface oxygen concentrations on thin and thick turf algae were hyperoxic (more than 500 µM) during the day and anoxic (less than 2 µM) during the night. Surface oxygen concentrations on A. rhodantha and H. majus were hypoxic or anoxic during both day- and night-time. Although these results may appear unexpected during day-time for these photosynthesizing organisms, surface measurements refer to measurements taken at a depth of 0 µm, which is the hard (skeletal) surface of the core, and not from the surface of the algae/mat itself (see Material and methods). Thus, these measurements were taken underneath the algal canopy or cyanobacterial mat where hypoxic conditions could occur throughout the day and the night. In contrast, surface oxygen concentrations on conspecific coral and CCA remained above 100 µM at night.

DBL thickness remained similar between day and night for each benthic competitor × sampling position combination (figure 2c,d). The interfaces with H. majus, A. rhodantha, and thick turf algae showed a thick DBL (approx. 3 000 µm). In contrast, the interfaces with thin turf, CCA, and conspecifics showed a thin DBL (approx. 500 µm). DBLs on competitor surfaces were thickest for H. majus, A. rhodantha, and thick turf algae and thinnest for conspecifics, CCA, and thin turf. DBL thickness on apparently healthy Porites tissue was always less than 1 000 µm. It was higher in interactions with A. rhodantha and H. majus, probably owing to the effect of their high canopies affecting flow regime downstream beyond the 1 cm distance separating the interface (i.e. edge of the canopies) and coral measuring positions.

Rates of overgrowth correlated significantly with night-time surface oxygen concentrations and DBL thickness (Spearman's rank-order correlations, r² = 0.889 and 0.785, respectively, p < 0.05) and not with day-time surface oxygen concentrations or diel oxygen fluctuations.

(c) Experiment 2

Day-time hyperoxia and night-time hypoxia at the interface were alleviated by increasing ambient flow, except for when thick turf was upstream (figure 3a,b, complete ANOVA results in electronic supplementary material, tables S2 and S3). Increased flow also reduced day-time surface oxygen concentrations on the turf algae, regardless of flow orientation. At night, thick turf remained anoxic, regardless of flow speed. Concentrations on apparently healthy Porites tissue during day or night remained stable, regardless of flow speed and direction.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

(a) Day- and (b) night-time surface oxygen concentrations, and (c) day-time DBL thickness as a function of flow speed (FS) and direction (D) in interactions between massive Porites spp., with thin and thick turf algae. Data points are mean ± s.e.m. (n = 4). See electronic supplementary material, tables S2 and S3 for complete statistical results. Letters indicate significant groupings of flow speeds for each of the three measuring positions (MP) by Tukey multiple comparison tests. Vertical lines indicate significant groupings of measuring positions within each flow speed. Algae–coral: algae upstream; coral–algae: coral upstream. Note scale differences on the y-axis for day and night.

DBL thickness showed similar trends between day and night, so only day-time values are presented (figure 3c, night values in electronic supplementary material, figure S3). DBL thickness at the interface decreased with increasing flow, except for when thick turf was upstream. When the coral was upstream, DBL thickness at the interface between coral and thick turf became rapidly similar to the DBL thickness on the coral surface with increasing flow. In contrast, when thick turf was upstream, values at the interface remained close to those present on the algae.

(d) Experiment 3

When ampicillin was added, night-time hypoxia was alleviated both at the interface and on the algae (figure 4, complete ANOVA results in electronic supplementary material, table S4). The reduction of hypoxia at the interface matched the reduction at the algal surface for both thin and thick turf. Surface oxygen concentrations on apparently healthy Porites tissue declined slightly with antibiotic addition.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Night-time surface oxygen concentrations for thin and thick turf in the presence and absence of antibiotic treatment (A−, without ampicillin; A+, with ampicillin, respectively). Turf algae were placed upstream of Porites. Data points are mean ± s.e.m. (n = 3). p-values are from rank-based ANOVAs. See electronic supplementary material, table S4 for complete ANOVA results. Letters indicate significant groupings for each of the three measuring positions (MP) by Tukey multiple comparison tests. Horizontal bars indicate significant groupings of measuring positions for each antibiotic (A) and turf category (TC) treatment combination.

4. Discussion

Benthic algae are commonly grouped into functional groups, including CCA, turf algae, and fleshy macroalgae [2,10]. CCA are typically poor competitors against corals and are even used as controls in studies of coral–algal interactions [31,40]. They are often associated with healthy reefs [8] as they facilitate the settlement and survival of recruiting corals [41] and prevent the colonization of macroalgae [42]. In contrast, fleshy macroalgae are frequently superior competitors against corals, inhibiting coral growth, reproduction, and recruitment [43–46]. Turf algae and benthic cyanobacteria have been less studied but they have also been shown to negatively affect corals at all life stages [12,32,47]. Our study confirms this ranking of competitiveness among algal functional groups against the massive coral Porites. In addition, we found that thin turf algae (less than 5 mm in height) did not overgrow corals as rapidly as thick turf algae (5–20 mm), highlighting the importance of turf height in coral–algal interactions. Elevated turf algae have been negatively associated with coral cover [48] and are known to be deleterious to coral settlement and survival [32,47].

Importantly, we show that the competitive abilities of the different functional groups, expressed as rate of overgrowth, matched patterns of hypoxia during the night and DBL thickness at the coral–algae interface, supporting (but not demonstrating) a causal link between the physiological processes occurring at the coral–algal interface and the competitive dynamics in the field. Coral interfaces with thick turf algae, macroalgae, and cyanobacteria were characterized by a thick DBL and strong hypoxia at night, whereas coral edges bordered by conspecifics, CCA, and thin turf had thin DBLs and less reduced oxygen concentrations at night. These results support the importance of water-mediated mechanisms in coral–algal interactions, in particular in interactions between corals and dense algal and cyanobacterial assemblages which are commonly encountered in degraded reefs. To date, a number of studies support the importance of allelochemical interactions among corals and algae. Seaweed extracts experimentally induce coral mortality in direct contact, with the extent of damage matching that caused by intact, live seaweeds [13,14]. Allelochemicals produced by algae are frequently hydrophobic and rely on transfer via direct contact. They are present on algal surfaces with highest concentrations found on basal portions of blades [15]. However, contact- and water-mediated mechanisms are not mutually exclusive and might act in synergy. It is plausible that allelochemicals transferred from algae to corals by direct contact initiate coral damage and subsequently facilitate the action of hydrophilic compounds.

Our results suggest that DBL thickness and/or hypoxia during the night at the coral–algal interface are important variables influencing coral tissue loss. A thick DBL at the coral–algal interface is likely to alter several physical, chemical, and microbial properties of the microenvironment. For example, mat-forming algal species, which severely restrict water exchange, decrease light and oxygen, and increase concentrations of DOC and soluble reactive phosphorus above understorey corals, and corals exposed to these conditions (i.e. pre-incubated seawater with macroalgae and shading) show significant physiological stress [49]. Likewise, a thick DBL is likely to facilitate the development of night-time hypoxia by impeding oxygen diffusion, limiting the removal of metabolic waste products, and enhancing the water-mediated transport of hydrophilic compounds (e.g. DOC) and microbes at the coral–algal interface [10]. Hypoxia, especially at night, limits coral respiration [25] and reduces the metabolism of coral symbionts [50]. Hypoxia has been measured in experimentally initiated and naturally occurring interactions between corals and turf or macroalgae [8,22,27,31,37]. Stressful pH levels are also likely to develop during the night in conjunction with hypoxia as a result of net respiration of turf or macroalgae [51]. A combination of locally reduced pH and oxygen can rapidly kill coral tissue [52], and a decrease in ambient pH values can facilitate coral overgrowth by algae [53].

In our study, high oxygen concentrations were measured at the coral–turf interface during daytime. Such hyperoxia has also been shown to create oxidative stress and bleaching in corals [26,54]. Hence, it cannot be ruled out as a stressor contributing to coral tissue loss, particularly in coral–turf interactions. However, we did not find a significant correlation between competitive gain/loss against the coral and day-time oxygen concentrations or diel oxygen fluctuations.

Alternatively, thick DBLs ameliorate the negative effects of ocean acidification by providing a biologically controlled buffer between the calcifying organism's external structure and the outer bulk seawater, potentially reducing rates of dissolution [55–57]. Dense algal canopies can decrease the susceptibility of understorey species to ocean acidification by providing regions of slow flow and higher day-time pH [57,58]. For example, reduced seawater velocities beneath the canopy-forming macroalga Carpophyllum maschalocarpum resulted in increased DBL thicknesses, higher pH (up to 8.9) and oxygen concentrations in the light, and lower pH (down to 7.74) and oxygen concentrations in the dark in understorey crustose coralline macroalgae [58]. Subsequent investigations should test the hypothesis that certain coral–algal interfaces could act as buffer zones against the corrosive effects of ocean acidification.

The oxygen concentrations at the coral–competitor interface largely matched those occurring on the algae, suggesting that algal-associated metabolites and microbes can spread to neighbouring corals, particularly when the algae are located upstream of the dominant current. When testing for the effect of flow speed on the coral–turf interactions, we found that increased flow reduces DBL thickness and oxygen extremes at the coral–algae interface. Increased flow commonly results in thinner DBLs [56,59]. Previous studies showed that high flow reduces algal competitiveness over corals and bacterial concentrations at the coral–algal interaction zone [60,61]. Furthermore, our study shows that thick turf oriented upstream retains the hyper- and hypoxic conditions at the coral–algal interface more strongly compared to thin turf, supporting the importance of algal canopy height, flow speed, and orientation in water-mediated interactions between corals and algae. Elevated turf algae are known to create thick DBLs [23]. They might create a thick DBL at the downstream interface with corals by acting as a physical barrier, especially in low flow conditions. In turn, this could facilitate the accumulation of water-soluble leachates and microbes from algae over downstream neighbouring corals. Here, we studied unidirectional flow. However, in situ interactions are likely to be influenced by oscillatory or bidirectional flows that reverse according to the wave phase, although there may be net flow created by waves of similar wavelengths travelling in the same direction [62]. More research is needed to investigate precisely to what extent such flows disrupt the DBL and alleviate diffusion limitation at the interaction zones.

Hypoxia at the interaction zones between corals and algae has been suggested to be the result of bacterial respiration, with the hypothesis that microbes cause highly hypoxic conditions at the coral–algal interface owing to tissue degradation [31]. In our study, antibiotics reduced the hypoxic conditions at the coral–turf interface, suggesting that hypoxia during night-time is partially due to bacterial respiration. Antibiotics also suppressed the hypoxic conditions on the turf. The microbial biofilm at the interface could be fed by DOM from the algae leaching downstream during the day and resulting in high rates of microbial metabolism during the night. In such a case, oxygen depletion as a result of coral tissue degradation may be relatively minor. In a similar flume experiment, Brown & Carpenter [27] found no evidence of hypoxia caused by bacteria at the interface between massive Porites and turf. Possible explanations include different species composition of the turf assemblage and lower antibiotic concentrations.

As reefs transition towards macroalgal dominance, their algal assemblages commonly change from coralline algae and thin turf to thick turf, macroalgae, and cyanobacterial mats. Our results provide evidence for the importance of water-mediated mechanisms in coral–algal interactions and suggest that these mechanisms will increase as reefs shift towards fleshy algal domination. Denser algal assemblages may lead to thicker DBLs at coral–algal interfaces. In turn, this may lead to higher frequencies of hypoxia and higher concentrations of harmful metabolites and pathogens along coral borders, which would ultimately facilitate algal overgrowth of live corals and produce a positive feedback loop for coral reef decline [4]. High water flow could alleviate these conditions, as well as other factors (e.g. herbivores, nutrients) that control algal abundance. For example, in low flow areas, grazing by herbivores could help decrease the height and density of turfs and the abundance of macroalgae [3,4,46,48], which in turn could mitigate the effects of water-mediated interactions between corals and algae by decreasing DBLs and preventing hypoxic conditions along coral edges.

Data accessibility

The datasets supporting this article can be found on Dryad doi:10.5061/dryad.t8j15.

Authors' contributions

H.J. collected and analysed the data and drafted the manuscript; C.S. collected the data; R.O. and D.B. participated in the design of the study and provided materials; M.N. designed the study, provided materials, and drafted the manuscript. All authors contributed to writing of the manuscript and gave final approval for publication.

Competing interests

We declare we have no competing interests.

Funding

Funding was provided by the CNRS Chaire d'Excellence to M.M.N.