Reviewing Evidence for Disturbance to Coral Reefs Increasing the Risk of Ciguatera

Abstract

The hypothesis that disturbance to coral reefs creates new surfaces that increase the risk of ciguatera is premised upon the increased algal substrates that develop on these surfaces being colonised by high ciguatoxin (CTX)-producing Gambierdiscus species that proliferate and enter the ciguatera food chain. Current evidence indicates that new algal substrates are indeed rapidly colonised by Gambierdiscus. However, the requirement that these Gambierdiscus species include at least one that is a significant (high) CTX-producer is more likely a limiting step. While ambient environmental conditions impact the capacity of Gambierdiscus to bloom, factors that limit the growth of the bloom could influence (typically increase) the flux of CTX entering marine food chains. Additionally, new algal substrates on damaged reefs can be preferentially grazed to funnel ciguatoxins from Gambierdiscus to herbivores in disturbed reef areas. In societies consuming second trophic level species (herbivores, grazers, and detritivores), such funnelling of CTX would increase the risk of ciguatera, although such risk would be partially offset over time by growth (toxin-dilution) and depuration. Here, we review evidence for six potential mechanisms to increase ciguatera risk from disturbance to coral reefs and suggest a hypothesis where ecosystem changes could increase the flux of CTX to groupers through a shift in predation from predominately feeding on planktonic-feeding prey to mostly feeding on benthic-feeding prey, increasing the potential for CTX to accumulate. Evidence for this hypothesis is stronger for the Pacific and Indian Oceans, and it may not apply to the Caribbean Sea/Atlantic Ocean.

1. Introduction

Ciguatera is caused by eating normally edible, marine, warm-water fishes contaminated with ciguatoxins (CTX) [1]. It results from a chain of events that begins with the production of CTX by benthic dinoflagellates (Gambierdiscus spp. and possibly Fukuyoa spp.), which are then transferred through marine food chains to produce fish that cause human poisoning, with some analogs structurally bio-modified in the process [2,3,4,5,6,7,8]. Ciguatera in the Pacific basin appears to be caused mostly by two structural families of toxins that can co-occur in contaminated fish, analogs of Pacific-ciguatoxin-1 (P-CTX-1, also known as CTX1B) and analogs of P-CTX3C (CTX3C) [3,9,10]. The USFDA has set a conservative recommendation of 0.01 µg/kg CTX equivalents (eq.) for the safe consumption of seafood, whereas Japan has recommended a less stringent safety limit of 0.175 µg/kg [11,12]. The USFDA limit is based upon the analysis of Lehane and Lewis [13], suggesting 0.1 µg P-CTX-1/kg flesh may cause mild poisoning in 2 out of 10 people eating 500 g of flesh (i.e., consumption of 50 ng P-CTX-1 eq.). The less stringent safety limit in Japan may partially reflect smaller portion sizes of fish meals than those typically consumed in other cultures. Ciguatera in the Atlantic and Indian Oceans is mostly caused by different structural families of CTX, Caribbean-CTX (C-CTX) in the Atlantic Ocean and Indian Ocean-CTX (I-CTX) in the Indian Ocean [3,14,15,16,17,18,19,20].

A long-held hypothesis originating from the Pacific basin is that disturbance to coral reefs causes an increased risk of ciguatera [21,22,23,24,25,26,27,28,29,30,31,32,33,34]. This hypothesis was initially derived from observations of apparently increased ciguatera risk after damage to Pacific coral reefs during and in the aftermath of World War II and before the discovery of Gambierdiscus as an origin of CTX, with ciguatera being a problem for both Japanese and US forces in Micronesia during the conflict [35]. In what became known as the “new surface hypothesis”, Randall [23] suggested that disturbance created new surfaces for the causative organism to colonise and proliferate. This hypothesis implies that space for growth of Gambierdiscus populations is limiting or that post-disturbance food chains increase the transfer of CTX to species consumed by people. However, if new surfaces can increase the risk of ciguatera, then some natural phenomena, including violent storms, should also create conditions that promote ciguatera outbreaks [25,36]. Fleshy macroalgae and/or turf algae can colonise and dominate coral reefs after major disturbances [37,38,39,40,41,42,43,44,45], providing new surfaces for Gambierdiscus to attach and possibly proliferate. However, there are few direct studies that have tested this hypothesis. Kaly and Jones [46] could not find supporting evidence for the new surface hypothesis from studies on reef disturbance at sites in Tuvalu, and engineering works for the development of a small marina site on a fringing reef at Hayman Island on the Great Barrier Reef in the 1980s did not produce an increase in Gambierdiscus populations [4]. In the only study of the toxicity of surgeonfishes (Ctenochaetus striatus) from the Great Barrier Reef, Lewis et al. [47] found only low concentrations of CTX in C. striatus spearfished from John Brewer and Davies reefs, despite John Brewer reef being damaged by crown-of-thorns starfish (Acanthaster sp.) while the nearby Davies Reef was only lightly impacted at the time. Thus, if disturbance to coral reefs leads to increased ciguatera risk, food chain factors likely play a significant role, given that most disturbances do not appear to be associated with outbreaks of ciguatera.

Turf algae rapidly colonise new surfaces on coral reefs [45,48,49,50,51,52,53] that are, in turn, likely colonised by Gambierdiscus transferred through the water column. This probably happens quickly, given that Gambierdiscus can be detected within 24 hr of deployment of benthic screen assays [54,55]. Several authors have advocated a greater focus on turf algae as substrates for Gambierdiscus as part of food chains that contaminate carnivorous fishes that most often cause ciguatera [2,4,6,7,56,57,58]. This may be more relevant for the Pacific Ocean, where many more herbivorous fish feed on turf algae than macroalgae [59,60,61,62], with the possibility that turf algal ecosystems will expand as coral reefs continue to deteriorate in the Anthropocene [62,63,64,65]. However, the difficulty in sampling and quantifying benthic dinoflagellate populations from algal turfs [47,66,67,68] has limited research into the role of turf algae in ciguatera. This could be addressed if test screen assays [54,55] or other technologies [47,66,67,68,69] are more widely deployed to quantify benthic dinoflagellate populations on turf algae. Although turf algae are well recognized as part of the mix of inorganic and organic substrates that make up large parts of coral reefs, these habitats are often excluded from analyses of substrate dominance on reefs, biasing perceptions of what typical substrates are on coral reefs [62,63,70]. It is also unclear if high CTX-producing species of Gambierdiscus differentially distribute on different algal substrates. These factors confound the use of published substrate data to quantify the effects of disturbance to coral reefs on ciguatera risk. However, quantitation of Gambierdiscus concentrations on turf algae combined with remote sensing techniques that estimate the spatial extent of turf algae in shallow waters (<20 m) of coral reefs [71] could allow the quantification of reef substrates and potentially the development of predictive ciguatera risk assessment models for reefs [4,6,7].

Many coral reef systems, including the Great Barrier Reef, have been impacted by an increasing range and frequency of major disturbances from cyclones, outbreaks of crown-of-thorns starfish, increased catchment runoff from more intense rain events and ongoing development, and coral bleaching events due to increasing water temperatures associated with climate change [65,72,73]. At least for the Great Barrier Reef, this has not yet produced any obvious increase in outbreaks of ciguatera [4], which supports the contention that not all disturbances impact ciguatera risk. However, disturbances cause different impacts to coral reefs and act over different time scales. For example, intense marine heat waves can produce dead but intact coral skeletons, whereas tropical storms pulverize and remove coral structures to leave behind a more open planar surface, leading to different outcomes for algal growth and herbivory. Dead coral skeletons can not only facilitate the establishment of algae but, in some circumstances, favour an algal assemblage that is more resistant to control by herbivores [74]. Biological disturbances such as outbreaks of crown-of-thorns starfish have limited influence on the structural complexity of reef habitats in the short term compared to the immediate physical damage from storms, although the longer-term impacts may be similar [75]. For outbreaks of ciguatera to increase, there must be an increase in the amount of CTX that accumulates in fish that are eaten by people or a shift in the pattern of human consumption to higher-risk fish species.

Changes that increase the risk of ciguatera could occur at each trophic level of marine food chains, or in the transfer processes between trophic levels, or combinations of these (Figure 1). At the base of the ciguatera food chain, these could occur through processes that:

• Increase cell numbers (a bloom) of resident CTX-producing Gambierdiscus, and/or,
• Increase cellular production of CTX in resident Gambierdiscus species or strains and/or,
• Shift the dominance of the species/strains of resident Gambierdiscus from low CTX-producing to higher CTX-producing populations (super-bug hypothesis).

Additionally, changes within existing food chains that could increase the amount of CTX transferred between trophic levels could occur through processes that cause:

4.

Behavioural changes in consumer species that ingest Gambierdiscus that increase their risk of predation and therefore increase the probability that CTX is transferred to higher trophic level fish more often consumed by people, and/or,

Ecosystem changes that act on food chains to:

5.

Alter the diet of herbivorous or carnivorous fishes that increases the flux of CTX transferred to higher trophic levels, and/or,

6.

Changes in abundance and/or size of reef fishes that alter the dynamics of reef food chains and the flux of CTX through them to human consumers. This could occur through natural processes such as variation in fish recruitment, and/or depletion of stocks from harvesting of marine resources.

In this paper, we review the evidence for these six potential mechanisms to increase ciguatera risk from disturbance to coral reefs and suggest a hypothesis where ecosystem changes (Section 2.5) could increase the flux of CTX to predatory fishes through a shortening of food chains that changes predation by groupers from prey species that predominately feed on planktonic organisms, to those that mostly feed on benthic organisms. However, the evidence for this hypothesis comes mostly from studies in the Pacific and Indian Oceans and may not apply to the Atlantic Ocean and adjacent waters. Our review assumes that Gambierdiscus and possibly Fukuyoa species are the principal origin of ciguatoxins and assumes other potential sources, such as cyanobacteria [76,77,78], contribute little to ciguatera risk [4].

2. Examination of the Six Potential Mechanisms to Increase Ciguatera Risk from Disturbance to Coral Reefs

2.1. Increase Cell Numbers (A Bloom) of Resident CTX-Producing Gambierdiscus

The proliferation of Gambierdiscus through cell growth is widely hypothesised to create an increased risk of ciguatera and is consistent with ciguatera appearing in the absence of any large-scale reef disturbance. There is a comprehensive literature on factors influencing the growth of Gambierdiscus, beginning with its culture and environmental response studies from the 1970s [79,80,81,82]. This research has only increased since the discovery in the 1990s that the genus consisted of more than one species [3,83]. To date, 19 species of Gambierdiscus have been recognized [3,84], with many of these species suggested capable of producing CTX analogs, although their cell concentrations vary considerably [3,16,17,85,86,87]. Therefore, early studies on Gambierdiscus growth and toxicity need to be interpreted cautiously as they assumed a monotypic genus (G. toxicus), with later studies showing differences across species [85]. For a Gambierdiscus bloom to increase the risk of ciguatera, the species proliferating must be a significant CTX producer, as growth of low-CTX producing species are unlikely to cause an increased risk of ciguatera [2,4,7,58,85,88,89,90]. The discovery of increasing numbers of Gambierdiscus species that produce different arrays and levels of CTX-analogs has parallels to research on Alexandrium spp. and the paralytic shellfish toxins they produce [91].

Gambierdiscus populations could bloom through an environmental change that increases cellular growth rates and/or through a reduction in herbivore grazing pressure [4,67]. As yet, there is no direct evidence of disturbance to coral reefs stimulating Gambierdiscus growth, although grazing pressure on algae can be reduced from increased sedimentation [92,93,94,95,96], which may be exacerbated by reef disturbance. Indirect evidence for an association between peak Gambierdiscus densities and disturbance was reported in French Polynesia, where Gambierdiscus population densities and bloom frequency increased after coral bleaching caused by elevated seawater temperatures [90]. Rongo and van Woesik [32] also found a correlation between reef disturbance and increased ciguatera risk in Rarotonga in the southern Cook Islands, but there was no associated analysis of Gambierdiscus populations. Disturbance of sediments and damage to reefs could release or resuspend bioavailable nutrients into the water column [97,98], stimulating Gambierdiscus growth, and this, along with the creation of new surfaces, have been suggestions for how reef disturbance could increase ciguatera risk. However, the impact of sedimentation on Gambierdiscus growth and toxin production has not been assessed. In the absence of physical disturbance, nutrients are often input from terrestrial runoff to inshore reefs, while on outer reefs, nutrient levels can be influenced by upwelling [99,100,101].

The creation of new surfaces on its own is unlikely to stimulate dinoflagellate growth but would provide increased area for growth of algal substrates (macroalgae, turf algae) for Gambierdiscus. However, these new surfaces would only lead to an increase in ciguatera risk if they were colonised by CTX-producing species that then proliferated and were then grazed by fish or invertebrates. Short-term increases in herbivore populations, including species often associated with causing ciguatera, such as the surgeonfish Ctenochaetus striatus and other acanthurids, can occur in response to increased filamentous algae on bleached corals, potentially through migration to feed on more abundant or preferred sources of food [102]. This short-term increase in localized herbivory could lead to increased accumulation of CTX into these herbivores, likely contributing to the patchy distribution of ciguatoxic fishes. In societies where herbivorous/detritivorous fishes are directly harvested for food, such as island nations in the Pacific Ocean, this could increase ciguatera risk [6]. Conversely, in areas such as the Great Barrier Reef, where herbivorous/detritivorous fishes are generally not harvested for food [103,104], any increased levels of CTX into the second trophic level may not always lead to an increased risk of ciguatera [4].

Cage experiments have shown that macroalgae on coral reefs can be limited by grazing [38], but there is little evidence that the spatial coverage of turf algae is similarly limited by grazing. Such limitation may be unlikely as turf algae can grow rapidly and colonise numerous substrates. New surfaces can be colonized by turf algae within weeks [48,49,105] and be dominant after six months [51]. While disturbance to coral reefs can create new surfaces for turf algae to proliferate that may favour increased grazing in the short term [102], any associated increase in sedimentation can change the structure and productivity of turfs such that herbivore grazing is suppressed [70]. In the Caribbean, overfishing of herbivores, along with eutrophication and the die-off of sea urchins and corals, has been associated with proliferation of macroalgae [106,107,108,109]. If the smaller population of herbivores created by overfishing reduced competition for algal substrates supporting Gambierdiscus populations, this could facilitate the funnelling of CTX into these now smaller populations of herbivores [4]. However, this mechanism is not necessarily associated with physical disturbance to the coral reef systems. The often-cited increase in macroalgae on disturbed coral reefs appears to be a feature of the Caribbean, with little evidence for a general increase on Indo-West Pacific reefs [52,110].

French Polynesia saw an increase in herbivores that controlled macroalgae that proliferated after reef disturbance [111]. This feedback response suggests increased grazing associated with increased food availability (macroalgae) [112]. Even though this was considered a rapid response, the changes were significant only over a time scale of years [111], which compounds the difficulty of trying to attribute causal relationships for ciguatera risk. In addition, the environmental outcomes from reef degradation resulting from interactions between corals, macroalgae, and fish (among other factors) are complex [43,113,114,115] that can vary depending upon herbivore/grazer biomass, herbivore size, and fish community structure [103,116,117], and possibly with the state (initial conditions) of the reef before the disturbance [118]. These complex interactions limit any analysis of the factors that are more likely to lead to increases in ciguatera risk, especially when underlying mechanisms may differ between regions.

2.2. Increase Cellular Production of CTX in Resident Gambierdiscus Species or Strains

The hypothesis that resident populations of Gambierdiscus can increase the production of CTX in response to an environmental stimulus is intuitively appealing as an explanation for the unpredictable nature of ciguatera outbreaks. The first experimental support for this hypothesis that specifically examined changes in presumed CTX concentrations in Gambierdiscus (cellular Na⁺-channel toxicity) was described by Sperr and Doucette [119], who reported enhanced ciguatoxicity under batch culture growth using differing N:P nutrient ratios. However, it is not clear how changes in nutrient ratios per se affect toxicity, especially if the cells are nutrient-replete, which may be likely in the exponential growth phase of batch culture, irrespective of which nutrient ratio the cells are growing under. Possibly, batch cultures with differing nutrient ratios may be more likely to affect toxin production as growth becomes limiting and cultures enter stationary growth phase. It is interesting that toxin production is often highest at low growth rates of Gambierdiscus, i.e., under conditions that limit cell division [119,120,121,122,123], including some strains grown under nitrogen-limiting conditions [120]. How these laboratory culture conditions relate to natural environmental changes that affect ciguatoxin production remains to be determined but may be inferred from the relative change in CTX concentrations in cultures of some strains of Gambierdiscus. The maximum relative change in CTX concentrations is ~2- to ~3-fold across many cultures [120,122,123,124]. Such variation needs to be factored into future models of the production and flow of CTX in food chains (see recent modelling by Holmes and Lewis [5,6] and Parsons et al. [7]). It is possible that a 2–3-fold change in cellular CTX concentration induced by environmental factors alone could influence ciguatera risk, either up or down, depending on the specific suite of environmental factors at play.

Nutrients can be released into the water column from the resuspension of sediments [97,98]. The type of disturbance impacting a reef will affect the quantity and duration of nutrient release, from the chronic, long-term effects of land-based agriculture to short-term pulses such as from storms or ongoing dredging. Episodic disturbance could lead to nutrient pulses that promote Gambierdiscus blooms that then increase in toxicity as nutrients become limiting [120] as ambient nutrient concentrations return to the low levels typical of coral reef waters [101]. This would correspond to a two-step process to increase CTX load on a reef.

Changes in other environmental parameters, such as salinity and temperature, have also been shown to vary CTX concentrations in Gambierdiscus [121,122], but physical disturbances to coral reefs are unlikely to significantly alter these parameters in the short term. Vacarizas et al. [122] did find the highest CTX production for G. carpenteri growing under low light conditions. The light (photon flux density) transmitted through the water column would likely be reduced by increased turbidity associated with reef disturbance, although this will depend in part on the type and duration of the disturbance. While reduction in light may affect CTX production of an existing bloom of Gambierdiscus, it may also limit nutrient-driven growth. The decrease in light would presumably need to be of considerable magnitude, as the light received by benthic organisms typically changes constantly throughout the day and varies with season, water depth, cloud cover, and turbidity, all of which affect the clarity of the water column, and Gambierdiscus can orientate itself within the light environment [125]. Most culture studies do not attempt to mimic this variability of the light environment in the wild, relying instead on continuous fluorescent or LED lighting with unchanging day-night photoperiods. It would be interesting to test whether changes in light alter the toxicity of cultures instead of the typical practice of acclimating them to new light regimes before analysing for toxicity. Any effect from a reduced light regime would only be relevant over limited time scales for the transfer of CTX to herbivorous fishes, given the inverse relationship between increasing turbidity and herbivorous fish abundance on coral reefs, likely because of the impact of reduced light on the biomass and nutritional value of turf algae to herbivorous fishes [126]. Ciguateric predatory fishes have been caught from deep water (>200 m) where light levels are low; however, the origin of the CTX was suggested to be Gambierdiscus from shallower depths [127]. As suggested by Chinain et al. [85], further research is required to understand how changes in environmental conditions can affect toxin production across different species and strains of Gambierdiscus.

The possibility of biotic factors influencing CTX production and structural modifications of toxins also needs to be considered. Wang et al. [128] reported that co-occurring species of bacteria could increase or decrease CTX production in Gambierdiscus. There is also evidence that Gambierdiscus can acquire nutrients through mixotrophy [129,130]. However, studies on whether heterotrophic nutrition and its relative contribution to homeostasis can affect production of CTX by Gambierdiscus are lacking. There is also an increasing literature on how certain grazers can induce toxin production in populations of planktonic diatoms and dinoflagellates [131,132,133,134,135,136,137]. However, we are not aware of any such studies on CTX production in Gambierdiscus or other benthic dinoflagellates. If grazing by fish or invertebrates could induce increased production or accumulation of CTX in adjacent (down current) populations of Gambierdiscus, it could increase the CTX load on a reef. Inducible defence mechanisms to deter grazing are common in plants [138]. However, any such response for benthic dinoflagellates would need to be rapidly inducible to provide time to act as a grazing deterrent. Additionally, any such inducible deterrent to herbivore foraging could also operate for the algal substrate as for its epiphytic dinoflagellates. However, it is not clear how such biotic mechanisms would relate to reef disturbance.

In an early attempt during the 1980s to examine if reef disturbance could directly stimulate CTX production by Gambierdiscus through a hypothesised CTX-inducing factor [88], we added seawater extracts of the ground-up tips of living staghorn coral (Acropora sp.) collected from Flinders Reef in south-east Queensland [139] to cultures of Gambierdiscus isolated from Flinders Reef and the Great Barrier Reef [89]. These were cultures from which we could not initially detect any CTX by mouse bioassay [89]. After one week, we harvested the spiked cultures but again could not detect any CTX by mouse bioassay (unpublished results), so we could not find any support for the hypothesised CTX-inducing factor. These experiments were carried out at a time when Gambierdiscus was assumed to consist of only one species (G. toxicus), so we cannot attribute the species used in our experiments. In summary, while individual factors arising from damage or disturbance to coral reefs could change environmental conditions to affect the growth of Gambierdiscus, there is insufficient evidence of any consistent processes enhancing the production of CTX in resident populations of Gambierdiscus species or strains.

2.3. Shift the Dominance of the Species/Strains of Resident Gambierdiscus from Low CTX-Producing to Higher CTX-Producing Populations (Super-Bug Hypothesis)

Many species of Gambierdiscus are thought to produce CTX, but the cell concentrations can vary many-fold [3,16,17,85,86,87,124]. The existence of what we called super-producing strains of Gambierdiscus [2,89,140,141] originated from the detection of large differences in CTX concentrations between cultured and wild Gambierdiscus with large differences subsequently reported between and within species [3,86,123,124,142]. This concept of super-producing species/strains has since been conceptualized as the “superbug” hypothesis [7,143]. The highest-known producers of CTX analogs (toxin cell quotas) are G. polynesiensis in the Pacific Ocean [87,123,124,144] and G. excentricus and G. silvae in the Atlantic [121,145,146,147,148,149]. These three species can have cell toxin quotas >1,000-fold higher than other species examined, including G. caribaeus and G. carolinianus [86,121,123,147]. This inter-species range of CTX concentrations is far greater than the ~2–3-fold range known from intra-species studies of CTX concentrations quantified across different growth phases of cultures [120,122,123,124].

Gambierdiscus populations have long been known to vary seasonally on coral reefs [7,66,86,90,138,150,151], although only recently has it been suggested that ciguatoxicity can also vary seasonally in an inverse relationship with cell abundance [151], possibly indicating greater toxicity under growth limiting conditions. The CTX-load of Gambierdiscus populations do not necessarily correspond to cell numbers [2,4,86,88,89,90,151,152], with CTX-production initially suggested to vary between strains [2,88,89,90] and later shown to vary between and within species [3,16,17,85,87,121,124,142,146]. Sites can host multiple Gambierdiscus species [86,143], with the species mix changing through time [143], so it is not unexpected that ciguatera risk, as it relates to CTX-production, can also change through time. It has long been known that different Gambierdiscus cultures isolated from the same site and at the same time can have very different CTX toxicities [89]. However, there is currently no evidence for disturbance to coral reefs favouring the proliferation of “superbug” species, which shift the dominance of low CTX-producers to higher CTX-producing species of Gambierdiscus.

2.4. Behavioural Changes in Consumer Species That Ingest Gambierdiscus That Increase Their Risk of Predation and Therefore Increase the Probability That CTX Is Transferred to Higher Trophic Level Fish More Often Consumed by People

Many groupers are opportunistic carnivores that prey on a variety of fish and invertebrate species [153,154,155], so if herbivores (fish or invertebrates) were intoxicated by CTX and/or other toxic-metabolites produced by benthic dinoflagellates when feeding on high populations of Gambierdiscus, this could provide a mechanism for facilitating predation on CTX-contaminated prey [2,4]. Two Caribbean surgeonfishes (Acanthurus bahianus and A. chirurgus) fed a gel diet containing a Gambierdiscus species became disorientated and lost equilibrium [156], behaviour that in the wild would greatly increase their chance of predation. However, Magnelia et al. [156] found that surgeonfish could acclimate to feeding on a fixed dose of Gambierdiscus, suggesting that the intoxicating effects and risk of opportunistic predation from carnivorous fish species may be greatest after they feed on a sudden population increase (bloom) of Gambierdiscus. This may be a mechanism by which occasional blooms of ciguatoxin-producing benthic dinoflagellates lead to the production of ciguateric predatory reef fish species (reviewed by Holmes et al. [4]).

Clausing et al. [8,157] found that juvenile spotted unicornfish (Naso brevirostris) did not display any signs of intoxication when fed a gel diet containing G. polynesiensis. This surgeonfish accumulates CTX in the wild at Nuka Hiva (Marquesas archipelago, French Polynesia), although locals have considered it an edible species [158] with juveniles and sub-adults feeding on benthic macroalgae whereas adults tend to feed on gelatinous zooplankton [8,159,160]. However, it is likely that the behavioural response of surgeonfishes to feeding on substrates supporting Gambierdiscus populations depends upon the species and size of the herbivore consuming them, the cellular concentration of CTX (or other metabolite(s) inducing signs of intoxication in the herbivore), and the number of cells being consumed [4,6]. The absence of abnormal swimming behaviour in dusky groupers (Epinephelus marginatus) fed naturally CTX-contaminated fish flesh [161] suggests that the induction of abnormal behaviour is species-specific or the result of fish (especially herbivores) consuming toxic metabolites from Gambierdiscus other than CTX. Modelling by us [6] suggests that in some circumstances, more Gambierdiscus cells may be consumed daily by surgeonfishes that feed on turf algae than in the experimental protocol by Clausing et al. [8,157]. However, the ability of models to predict CTX transfer through trophic levels is constrained by a lack of experimental data [4,5,6,7,8], with more experiments like those of Clausing et al. [8,157] needed to produce better models. Currently, behavioural changes induced by grazers/herbivores feeding on toxic metabolites produced by benthic dinoflagellate cannot be directly linked to disturbance to coral reefs.

2.5. Change in the Diet of Herbivorous or Carnivorous Fishes That Increases the Flux of CTX Transferred to Higher Trophic Levels

Groupers (such as species of Plectropomus, and Epinephelus) are carnivorous Serranid fishes targeted by fishers for food that also often cause ciguatera poisoning [3,4]. On Queensland’s Great Barrier Reef, the common coral trout (P. leopardus) is the focus of the commercial fin-fish fishery [162] and is also one the principal causes of ciguatera in Queensland, Australia [1,6,163]. While groupers tend to be opportunistic predators, different species typically prey on benthic and planktonic species to varying degrees. For example, P. leopardus predominantly prey on plankton-feeding fish on the Great Barrier Reef [154,155,164,165] as well as other mesopredators [166]. In contrast, barcheek coral trout (P. maculatus), which tend to be found on more inshore reefs than P. leopardus, often consume a greater proportion of benthic prey [155,167], although there is considerable overlap in the distributions of these two species [168], and they can hybridize [169]. Plankton-based diets are unlikely to be a major source for the transfer of CTX to carnivores, although the rapid colonisation of benthic screen assays [54,55] indicates that at least some Gambierdiscus are tycoplanktonic, so they possibly could be consumed by zooplankton. Zooplankton are thought to be the principal diet of flying fishes [170], and the recent finding of CTX in glider flying fish (Cheilopogon atrisignis) [171] possibly suggests that tycoplanktonic Gambierdiscus could be a secondary food chain for the transfer of CTX. As the major CTX found in the flying fish was the less-oxidised P-CTX-2, it supports the contention that the toxin was consumed from an organism close to the base of the ciguateric food chain [4]. Intuitively, grouper species that principally prey on plankton-feeding fishes should be less likely to accumulate CTX. However, behavioural changes in potential prey species (reviewed in Section 2.4) could be a mechanism that increases the flow of CTX into opportunistic carnivores, including P. leopardus [4,6]. Recent studies in the Maldives (Indian Ocean) on groupers (Serranids) and tropical snappers (Lutjanids) from coral reefs support their reliance on planktonic prey [172].

Reefs around the Keppel Island group on the southern Great Barrier Reef were typified by exceptionally high coral cover, but much of it was lost due to coral bleaching and increased inundation by sediment-laden, freshwater flood plumes [173]. Long-term monitoring of these reefs demonstrated that, as coral cover declined, there was a decrease in the biomass of prey for mesopredators such as groupers, and a shift in dominant prey species from pelagic, plankton-feeding damselfishes to territorial benthic, algal-feeding damselfishes [173]. This resulted in an effective shortening of the food chain for P. maculatus from longer planktonic dominant food chains to shorter benthic dominant food chains [173]. Such a switch would suggest that these predators would be feeding on prey that had a greater likelihood to have ingested epiphytic dinoflagellates such as Gambierdiscus and Fukuyoa and, therefore, more likely to accumulate CTX if it was being produced by the local benthic dinoflagellates. A similar shortening of food chains, with a switch to greater feeding on benthic-feeding prey, was also reported for the grouper Cephalopholis argus after disturbance to reefs in the Seychelles, Indian Ocean [174]. These findings suggest that this may be a mechanism that can increase the risk for CTX to enter the food chain of mesopredatory fishes that are the focus of many reef fisheries. Several other studies in the Pacific and Indian Oceans have also reported reductions in plankton-feeding fishes after disturbance to coral reefs [175,176,177], indicating that predators could be forced to seek alternate, benthic sources of nutrition. This shift in predation could lead to an increased flux of CTX to higher trophic levels without any increase in the production of CTX or in the toxicity of herbivores.

The shift in dominant prey species for groupers from planktonic-feeding damselfishes to benthic-feeding species [173] suggests an avenue for future research for the food chain transfer of CTX. To date, much ciguatera research has focussed on surgeonfishes as the vectors for the transfer of CTX from benthic dinoflagellate source to carnivorous fish such as groupers (reviewed by Holmes et al. [4]). However, early reports from the Pacific included damselfishes among ciguateric species [21,22,178,179]. Additionally, the omnivorous damselfish Stegastes diencaeus eats benthic microalgae and crustaceans [180] and has tested positive for CTX in the Caribbean [181]. It has also been suggested that the loss of larger-bodied herbivorous fishes can lead to increased biomass and abundance of algal-farming damselfishes on reefs [108], which could possibly increase the flux of CTX into food chains.

Morillo-Velarde et al. [61] reported changes in trophic pathways for predatory fishes on degraded reefs in the Caribbean, but unlike the studies of Hempson et al. [173,174], there was no shortening of food chain length. Also, in contrast to the studies on degraded reefs in the Pacific and Indian Oceans, on degraded Caribbean reefs, the major carbon source for the carnivorous fish species studied changed from benthic sources (especially turf algae and epiphytes) to particulate organic matter that would likely include a greater planktonic component. In this case, it would suggest that disturbance to coral reefs in the Caribbean may not lead to an increased risk of CTX being transferred along food chains into groupers, as hypothesised for the Pacific and Indian Oceans. The limited studies with contrasting conclusions from the Caribbean Sea compared to the Pacific and Indian Oceans indicate the need for more research and raises the possibility that there may be different processes controlling the flow of CTX along food chains in different geographic regions. However, it would not be surprising if there were differences between the Pacific and Caribbean pathways for CTX accumulation into higher trophic level fishes on coral reefs, as their benthic ecosystems are reported to function in fundamentally different ways [52,64,182]. This has arisen in part because of their different biogeographical histories, which has led to a more than 3-fold difference in the diversity of fishes between the Indo-West Pacific and Caribbean regions [52,182,183] and in how different fish assemblages use reef and adjacent habitats [184]. This affects the ecosystem functioning of herbivory, with the diversity of herbivorous fishes of the Western Atlantic considered depauperate compared with the Indo-West Pacific [52,185,186]. Thus, we expect that the mechanisms controlling the development of ciguateric fishes will vary between regions.

The impacts arising from the biogeographical separation of tropical waters between the major oceans on how ciguatera and ciguatoxicity develops across regions is a matter of debate. Experimental evidence and recent modelling suggest that only high-CTX-producing Gambierdiscus species pose a major risk for producing ciguateric fishes capable of poisoning people, especially for P-CTX-1 analogs [2,4,5,6,7,8,58,85,88,89,90,157]. Currently, high-CTX-producing species of Gambierdiscus are only known from the Pacific and Atlantic Oceans but can be reasonably inferred to also occur in the Indian Ocean. The tropical waters of the Pacific and Atlantic Oceans have been geographically separated for ~3 million years by the Isthmus of Panama [187]. In contrast, the flow of tropical water between the Pacific and Indian Oceans may not have been as constrained by land barriers as sea levels changed during ice ages. Even during the last glacial maximum (~20,000 years ago), tropical waters of these two oceans were likely still connected via straits between the Sunda Shelf (encompassing much of Southeast Asia) and Sahul (Australia, Papua New Guinea) [188,189]. Thus, a connection potentially allowing the transfer of Gambierdiscus on floating debris [190] or by tycoplanktonic cells transported incrementally on tidal currents via a stepping-stone process [191] may have been possible between these two oceans for considerably longer than between the tropical Pacific and Caribbean. However, many of the same Gambierdiscus species are known from both the Pacific and Atlantic Oceans [3], suggesting limited endemicity among this group. It is interesting to speculate on the extent to which biogeographical differences have driven the biochemistry of CTX analogs between and across different oceans. Loeffler et al. [192] have suggested the presence of P-CTX3C in fishes from the Indian Ocean (previously, this analog has been only known from the Pacific Ocean), although their result requires confirmation given the uncertainty in tracing the origin of contaminated fish frozen and stored for ~3 years after capture and processing. Similarly, putative analogs of I-CTX have been recently detected from groupers caught from Okinawa that were also contaminated with P-CTX-1 [193]. Analysing the population genetics of fish to confirm the region of capture as well as the CTX-analogs produced by Gambierdiscus isolated from reefs in waters that have had long-term connections between the tropical Pacific and Indian oceans will help confirm that the distribution of P- and I-CTX analogs are indeed not restricted to one Ocean. Based upon maps of changing shorelines over the Pleistocene epoch [189], this could include reefs in and around the Banda and Savu Seas, adjacent to Indonesia and Timor Leste. This area is part of the Indonesian Throughflow, a series of currents that form the only tropical pathway connecting the global oceans and that have a crucial role in heat and water exchange between the Pacific and Indian Oceans [194,195]. As this drives water through Indonesia into the Indian Ocean it could provide a transport pathway for Gambierdiscus from the Pacific into the northeastern Indian Ocean. We are unaware of any studies of Gambierdiscus from this region; see recent review of ciguatera in the Indian Ocean [196]. However, benthic reef fish contaminated with P-CTX-1 are known to occur in the Arafura Sea [197], adjacent to the Indonesian Throughflow. We therefore predict that P-CTX-1-producing dinoflagellates likely occur in the northeastern Indian Ocean.

2.6. Changes in Abundance and/or Size of Reef Fishes That Alter the Dynamics of Reef Food Chains and the Flux of CTX Through Them to Human Consumers: This Could Occur Through Natural Processes Such as Variation in Fish Recruitment and/or Depletion of Stocks from Harvesting of Marine Resources

The conceptual model for the origin of ciguatera is based on a simplistic, linear model where Gambierdiscus blooms are incidentally consumed by herbivorous fish, which are preyed on by carnivorous fish, that are caught and eaten by humans, who are then poisoned. This is because the analysis of the role that ecological processes play in the production of CTXs and their transfer/bioconversion across trophic levels to cause ciguatera is still in its infancy and mostly relies upon data from ecological studies not directly related to ciguatera. We only use discrete trophic levels to describe the transfer of CTX along food chains in this review as our focus is on the transfer of toxins from source (Gambierdiscus) through intermediate vectors (invertebrate and/or herbivore), to predators. The estimation of trophic levels using stable isotopes integrates the assimilation of energy or mass flow through all the different trophic pathways leading to the organism [198], which produces non-integer trophic level numbers that have been recently used to estimate trophic magnification factors for CTX in fish [171]. However, this may not be representative of toxin transfer dynamics as the integrated sources of energy ingested can overwhelm those corresponding to the subset associated with toxin transfer, with δ¹⁵N and δ¹³C potentially reflecting the major nutritional sources rather than those associated with the production and transfer of CTX. For example, an initial vector in the CTX transfer process, such as the surgeonfish Ctenochaetus striatus, can have an equal or higher trophic level number than some groupers (Epinephelus maculatus, Variola albimarginata), their potential predators [171]. Ctenochaetus striatus is a detritivore that feeds by using its mouthparts to brush detritus aggregates from turf algae [95,199,200] and, in the process, ingests Gambierdiscus (reviewed by Holmes et al. [4]). The higher trophic number for C. striatus and some other herbivores [171] may occur because the substrate could include small invertebrates and material decomposing (recycled) from higher trophic levels, including fish faeces [199,201]. It is possible that the aggregated detrital material could also contain CTX, but this would be a different food chain pathway from the direct assimilation of CTX from Gambierdiscus.

The abundance and size distribution of grazers and predators in the ciguateric food chain are likely key factors affecting the transfer of CTX along food chains that lead to ciguateric fishes. Fish abundance can change considerably through natural processes such as variability in annual fish recruitment and predation, as well as from the human pressures of habitat modification, climate change, and fishing. The life stage at which fish accumulate CTX, as well as the frequency of any subsequent accumulations, are also likely important factors influencing whether sufficient toxin is transferred through trophic levels to produce a fish that can poison people. Additionally, dilution of toxicity through fish growth and depuration can reduce toxin concentrations in fish, although modelling suggests this may not occur quickly enough to reduce the risk of ciguatera from adult fish (reviewed by Holmes et al. [4]) with evidence of very rapid rates of depuration in juvenile fish [202,203].

In the Caribbean, long-term disturbance to coral reefs has resulted partly from the die-off of corals and sea urchins from disease and the effects of overfishing [106]. Parrotfishes (Labridae, Scarinae) are thought to contribute to the control of macroalgae, although some studies suggest that many are microphages [204,205] that may be more controlled by the benthos through bottom-up processes rather than exerting top-down control of macroalgae on coral reefs [206]. Parrotfish are targeted for food throughout much of the Caribbean, in part because of their greater availability relative to depleted stocks of higher-value large predatory fishes such as groupers and tropical snappers [207]. It is the larger-sized fish that contribute disproportionately to the control of macroalgae in the Caribbean if modelling by Bozec et al. [208] is correct in concluding that even modest (>10%) harvesting of large (>30 cm) parrotfish can reduce the recovery and resilience of Caribbean reefs. If the combination of overfishing and increased macroalgal biomass increases the risk of ciguatera, it would be reasonable to expect an increase in the rate of ciguatera poisoning to have occurred after the general increase in macroalgae in the Caribbean (assuming the relative impact of all other factors remained unchanged). This would not necessarily correspond to a uniform rate of increase, as the incidence of ciguatera varies greatly across the region [209]. However, we are not aware of any studies reporting significant increases in ciguatera occurring after the increase in macroalgae in the Caribbean, although there are suggestions of increases associated with warming sea surface temperatures [210] and increased storm frequency [33].

In the Pacific, grazing pressure increases non-linearly with increasing grazer biomass [211], although grazing pressure for a given biomass tends to be greatest with smaller fish sizes [212]. So, targeted fishing that removes larger-sized herbivorous fish may increase overall grazing pressure if the relaxation of competition from larger-sized herbivores leads to a relative increase in the cohort of smaller-sized herbivores. Although grazing pressure appears to be greatest when the population structure of larger-sized grazers is truncated [212], the ecological processes of herbivory vary significantly across the Pacific [211]. It has even been suggested that a population structure with smaller-sized parrotfish could counterintuitively favour macroalgal growth through the greater removal of epibionts from the surface of macroalgae [213].

In French Polynesia, high fishery value combined with slow life histories predispose unicornfishes (Acanthuridae, surgeonfishes) to overexploitation compared with parrotfishes [116]. If overfishing of high-value unicornfishes led to an increase in macroalgae then the increased substrate could allow for larger populations of Gambierdiscus to be consumed by the smaller remaining population of unicornfishes [4,6]. This provides a mechanism for the development of ciguatera, especially if compensatory processes do not operate among surgeonfishes and parrotfishes in the Pacific [211]. That is, the reduction in rates of herbivory from the loss of one type of herbivore is not offset by increases from another. Stocks of surgeonfishes (including unicornfishes) on the Great Barrier Reef are not directly affected by fishing as they have little commercial value for human consumption [103,104], and aggregate pressure from subsistence fishing on the Great Barrier Reef has been minimal. This is likely a contributing factor to the lower incidence of ciguatera from Australia compared to French Polynesia. The minimal commercial harvesting (and discouragement of recreational harvesting) of parrotfishes on the Great Barrier Reef may be an additional factor that reduces the risk of ciguatera on the east Australian coast relative to the Caribbean and Pacific Island nations.

While overfishing is usually related to human population pressures along with access to technologies that increase efficiency, the outcomes for any ecosystem are also dependant on cultural norms for consumption of seafood (that also changes over time). In many parts of the Pacific, overfishing of all trophic levels has occurred, with depletion of large-bodied predators taking place along with, or followed by, removal of large-bodied herbivores, especially macroalgal browsers [214]. This can lead to a growing dominance of small-bodied herbivores on coral reefs [215,216,217]. Interestingly, the removal of large browsers also favours increased biomass and abundance of algal farming damselfish [108], which could be a mechanism for the transfer of CTX to the remaining higher trophic level predators [4]. While overfishing has a long-term impact on coral reefs, it is not clear if disturbances to coral reefs can interact with extractive fishing to exacerbate ciguatera. The role of small herbivores as potential vectors for CTX transfer requires further investigation.

3. Conclusions

The provision of new surfaces after damage to coral reefs remains a viable hypothesis for increased ciguatera risk if the algae subsequently colonising the damaged reef are, in turn, colonised by a Gambierdiscus species that produce significant levels of CTX. Given how quickly benthic screens are colonised by Gambierdiscus [54,55], it seems likely that the new algal substrates will almost certainly be rapidly colonised by Gambierdiscus species. However, the requirement that these colonising Gambierdiscus species include at least one significant CTX-producer may limit which Gambierdiscus blooms increase ciguatera risk on reefs. In what appears common to both the Pacific and Atlantic Oceans, multiple Gambierdiscus species often co-occur on reefs, but the population densities of high toxin producers are frequently low [86,143]. The next step in the food chain to produce a ciguatoxic fish requires the proliferation of high-CTX-producing species or strains of Gambierdiscus growing on the new substrate. While the ambient environmental conditions likely control the capacity for Gambierdiscus to bloom, the CTX load on a reef may be influenced by factors that limit growth but increase cellular concentrations of CTX, with grazing pressure potentially limiting the capacity for Gambierdiscus populations to bloom [4,6].

For an increased risk of ciguatera, the substrate supporting a bloom of high CTX-producing Gambierdiscus should be preferentially consumed and CTX transferred to subsequent trophic levels. Indeed, there is evidence that new algal substrates on damaged reefs can be preferentially grazed [102,113]. In societies that consume herbivores, grazers and/or detritivores, this could increase the risk of ciguatera, although such risk would be partially offset over time by fish growth (toxin-dilution) and depuration [4]. Where carnivorous fish species are the main cause of ciguatera, CTX must be transferred from herbivores/grazers/detritivores to carnivores, with the number and toxicity of second trophic level fish greatly affecting CTX levels in the carnivorous fishes that ultimately poison people [5,6]. There is also evidence that disturbance to coral reefs in the Pacific and Indian Oceans can reduce the availability of prey species that predominantly feed on planktonic food sources, resulting in a shift to benthic-feeding prey for groupers [173,174], which are more likely to be contaminated with CTX on reefs supporting CTX-producing populations of Gambierdiscus.

Despite progress, more research is needed to understand the ecological processes that generate ciguateric fishes on both undisturbed and disturbed reefs. Could the rapid recovery of coral cover (but not necessarily function) that often occurs after damage to reefs [218,219] play a role in limiting ciguatera outbreaks? The proportion of toxic fish at Muroroa Atoll (French Polynesia) remains high (87%) despite nearly 30 years of recovery from the damage caused by nuclear testing [86]. It seems obvious that such tests would have caused significant damage to the surrounding reef, but although the last nuclear test at Muroroa Atoll was in late 1995, the Gambierdiscus species so far found there have limited ability to produce CTX (G. australes and G. pacificus) [86]. The difficulty in identifying the source of the benthic CTX load supporting the high proportion of toxic fish at Muroroa atoll or the ecological drivers maintaining toxicity highlights the challenges of understanding the drivers of increased ciguatera risk. We propose that key ecological processes controlling the production of ciguateric fishes remain to be elucidated, with our review suggesting that these likely vary at a regional level. As the world undergoes its fourth global coral bleaching event on record [220], it is important to better understand the key drivers affecting ciguatera risk that operate at these regional levels to mitigate the impact of ciguatera on low-income nations that rely so heavily on coral reefs for revenue and nutrition.